Benchmark run run-20260924-agy-descriptive
This page contains exactly the publishable benchmark observations recorded in run run-20260924-agy-descriptive; it does not select or aggregate them. Return to the run index or view aggregate metrics.
Interactive browser
Rows are pre-aggregated by task, target, prompt variant, MCP-instruction mode, context, arm, and restriction. Numeric values show median (p50), IQR, p90, maximum, and sample count. Boolean checks show pass counts and success rates. Filters apply to these summary groups; per-run observations remain available on their run pages.
Loading benchmark results…
Baseline vs semedit
Comparison evaluates baseline and semedit values within each restriction policy, target, and context group. Negative changes mark improvements for these lower-is-better metrics; positive changes mark degradations.
Detailed observations
Empirical telemetry measuring wall-clock latency, token usage, tool turns, and compiler correctness when AI coding agents perform code modifications.
Every benchmark pairs Vanilla LLM (standard file editing tools) against Semedit MCP (deterministic AST compiler operations).
Benchmark Methodology & Transparency
- Reproducibility & Provenance: Self-contained txtar archives define test scenarios. Links below resolve to commit-anchored GitHub source files for published commits, or cryptographic SHA-256 fingerprints for uncommitted local fixtures.
- Model Cost: Tables report model-specific credit consumption where a target model has a declared schedule: (uncached input × input credits + cached input × cached credits + (reasoning + visible output) × output credits) / 1,000,000. Thinking tokens use the output rate. ADR-0043 defines the per-million-token credit rates.
- Multi-Level Correctness Oracle: Four validation levels evaluate each trial:
- Level 1 (Mutation Policy): Restricts file modifications strictly to authorized paths.
- Level 2 (AST Invariants): Compiler AST verification of required symbols, imports, and relative declaration ordering.
- Level 3 (Clean Build): Strict compilation validation with zero compiler or typecheck errors.
- Level 4 (Test Suite): Unit and integration test suite execution.
- Isolated Sandboxing: Each trial executes in an ephemeral directory with pristine git and module state.
Test case: task-04-insert-public
Target: agy/gemini-3.8-flash/high
default prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-04-insert-public |
| Target | agy/gemini-3.8-flash/high |
| Prompt variant | default |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/bench/task_04_insert_public.txtar |
Vanilla LLM Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Add public constructor func InitServer() *Server placed before private helpers. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Add public constructor func InitServer() *Server placed before private helpers. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package api
type Server struct {
Port int
}
func (s *Server) Start() {}
func (s *Server) internalRun() {}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 87.40s | 102.44s | +17.2% | 68.83s | 129.05s | +87.5% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 21 | 28 | +33.3% | 14 | 33 | +135.7% |
| Initial Load / Discovery Turns | 14 | 19 | +35.7% | 6 | 12 | +100.0% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 20 | 27 | +35.0% | 13 | 32 | +146.2% |
| Output Tokens | 5063 | 7232 | +42.8% | 4030 | 9923 | +146.2% |
| Reasoning / Thinking Tokens | 3100 | 4363 | +40.7% | 2512 | 6823 | +171.6% |
| Total Input Tokens | 133716 | 238463 | +78.3% | 131931 | 334399 | +153.5% |
| Cached Input Tokens | 353149 | 552302 | +56.4% | 178817 | 649560 | +263.3% |
| Uncached Input Tokens | 133716 | 238463 | +78.3% | 131931 | 334399 | +153.5% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 2.64:1 | 2.32:1 | -12.3% | 1.36:1 | 1.94:1 | +43.3% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File api/server.go modified (9 lines -> 13 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File api/server.go modified (9 lines -> 13 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | view_file (✓ ✓) | view_file (✓ ✓) |
| 4 | view_file (✓ ✓) | view_file (✓ ✓) |
| 5 | run_command (✓ ✓) | run_command (✓ ✓) |
| 6 | run_command (✓ ✓) | view_file (✓ ✓) |
| 7 | view_file (✓ ✓) | view_file (✓ ✓) |
| 8 | run_command (✓ ✓) | view_file (✓ ✗): Created At: 2026-09-24T18:49:45+02:00 Completed At: 2026-09-24T18:49:45+02:00 Encountered error in step execution: declaring permissions: cortex tool view_file: convert tool call for permissions: model output error: invalid tool call error (invalid_args) failed to read file: stat /Users/alessandro/.gemini/antigravity-cli/mcp/semedit/instructions.md: no such file or directory |
| 9 | view_file (✓ ✓) | run_command (✓ ✓) |
| 10 | view_file (✓ ✓) | run_command (✓ ✓) |
| 11 | run_command (✓ ✓) | run_command (✓ ✓) |
| 12 | view_file (✓ ✓) | view_file (✓ ✓) |
| 13 | run_command (✓ ✓) | run_command (✓ ✓) |
| 14 | view_file (✓ ✓) | run_command (✓ ✓) |
| 15 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 16 | view_file (✓ ✓) | view_file (✓ ✓) |
| 17 | run_command (✓ ✓) | view_file (✓ ✓) |
| 18 | run_command (✓ ✓) | run_command (✓ ✓) |
| 19 | run_command (✓ ✓) | view_file (✓ ✓) |
| 20 | run_command (✓ ✓) | semedit/semantic_insert_function (✓ ✓) |
| 21 | — | view_file (✓ ✓) |
| 22 | — | run_command (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
| 24 | — | view_file (✓ ✓) |
| 25 | — | view_file (✓ ✓) |
| 26 | — | view_file (✓ ✓) |
| 27 | — | run_command (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File api/server.go modified (9 lines -> 13 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File api/server.go modified (9 lines -> 13 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | view_file (✓ ✓) | run_command (✓ ✓) |
| 3 | view_file (✓ ✓) | view_file (✓ ✓) |
| 4 | run_command (✓ ✓) | run_command (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | run_command (✓ ✓) |
| 7 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 8 | run_command (✓ ✓) | view_file (✓ ✓) |
| 9 | run_command (✓ ✓) | view_file (✓ ✓) |
| 10 | view_file (✓ ✓) | run_command (✓ ✓) |
| 11 | view_file (✓ ✓) | run_command (✓ ✓) |
| 12 | view_file (✓ ✓) | view_file (✓ ✓) |
| 13 | run_command (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 14 | — | run_command (✓ ✓) |
| 15 | — | view_file (✓ ✓) |
| 16 | — | view_file (✓ ✓) |
| 17 | — | view_file (✓ ✓) |
| 18 | — | run_command (✓ ✓) |
| 19 | — | view_file (✓ ✓) |
| 20 | — | view_file (✓ ✓) |
| 21 | — | view_file (✓ ✓) |
| 22 | — | view_file (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
| 24 | — | view_file (✓ ✓) |
| 25 | — | view_file (✓ ✓) |
| 26 | — | view_file (✓ ✓) |
| 27 | — | semedit/semantic_insert_function (✓ ✓) |
| 28 | — | view_file (✓ ✓) |
| 29 | — | run_command (✓ ✓) |
| 30 | — | run_command (✓ ✓) |
| 31 | — | run_command (✓ ✓) |
| 32 | — | run_command (✓ ✓) |
Target: agy/gemini-3.8-flash/medium
default prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-04-insert-public |
| Target | agy/gemini-3.8-flash/medium |
| Prompt variant | default |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/bench/task_04_insert_public.txtar |
Vanilla LLM Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Add public constructor func InitServer() *Server placed before private helpers. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Add public constructor func InitServer() *Server placed before private helpers. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package api
type Server struct {
Port int
}
func (s *Server) Start() {}
func (s *Server) internalRun() {}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 32.70s | 73.94s | +126.1% | 63.33s | 151.31s | +138.9% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 9 | 18 | +100.0% | 15 | 31 | +106.7% |
| Initial Load / Discovery Turns | 4 | 7 | +75.0% | 8 | 22 | +175.0% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 8 | 17 | +112.5% | 14 | 30 | +114.3% |
| Output Tokens | 2395 | 4500 | +87.9% | 3642 | 7142 | +96.1% |
| Reasoning / Thinking Tokens | 1457 | 2538 | +74.2% | 2042 | 4070 | +99.3% |
| Total Input Tokens | 85453 | 191337 | +123.9% | 111574 | 273960 | +145.5% |
| Cached Input Tokens | 93511 | 264222 | +182.6% | 211256 | 621148 | +194.0% |
| Uncached Input Tokens | 85453 | 191337 | +123.9% | 111574 | 273960 | +145.5% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 1.09:1 | 1.38:1 | +26.2% | 1.89:1 | 2.27:1 | +19.7% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File api/server.go modified (9 lines -> 13 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File api/server.go modified (9 lines -> 13 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | view_file (✓ ✓) | view_file (✓ ✓) |
| 3 | view_file (✓ ✓) | view_file (✓ ✓) |
| 4 | run_command (✓ ✓) | run_command (✓ ✓) |
| 5 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 6 | run_command (✓ ✓) | run_command (✓ ✓) |
| 7 | run_command (✓ ✓) | run_command (✓ ✓) |
| 8 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✗): Created At: 2026-09-24T18:46:18+02:00 Completed At: 2026-09-24T18:46:18+02:00 Encountered error in step execution: param "symbol" is required: invalid operation parameters |
| 9 | — | run_command (✓ ✓) |
| 10 | — | view_file (✓ ✓) |
| 11 | — | run_command (✓ ✓) |
| 12 | — | semedit/semantic_insert_function (✓ ✓) |
| 13 | — | view_file (✓ ✓) |
| 14 | — | run_command (✓ ✓) |
| 15 | — | run_command (✓ ✓) |
| 16 | — | run_command (✓ ✓) |
| 17 | — | run_command (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File api/server.go modified (9 lines -> 13 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File api/server.go modified (9 lines -> 13 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | view_file (✓ ✓) | view_file (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | view_file (✓ ✓) | run_command (✓ ✓) |
| 6 | run_command (✓ ✓) | run_command (✓ ✓) |
| 7 | run_command (✓ ✓) | view_file (✓ ✓) |
| 8 | view_file (✓ ✓) | run_command (✓ ✓) |
| 9 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 10 | run_command (✓ ✓) | view_file (✓ ✓) |
| 11 | run_command (✓ ✓) | view_file (✓ ✓) |
| 12 | view_file (✓ ✓) | run_command (✓ ✓) |
| 13 | run_command (✓ ✓) | run_command (✓ ✓) |
| 14 | run_command (✓ ✓) | view_file (✓ ✓) |
| 15 | — | run_command (✓ ✓) |
| 16 | — | view_file (✓ ✓) |
| 17 | — | run_command (✓ ✓) |
| 18 | — | view_file (✓ ✓) |
| 19 | — | run_command (✓ ✓) |
| 20 | — | view_file (✓ ✓) |
| 21 | — | view_file (✓ ✓) |
| 22 | — | view_file (✓ ✓) |
| 23 | — | semedit/semantic_insert_declaration (✓ ✓) |
| 24 | — | view_file (✓ ✓) |
| 25 | — | run_command (✓ ✓) |
| 26 | — | run_command (✓ ✓) |
| 27 | — | run_command (✓ ✓) |
| 28 | — | view_file (✓ ✓) |
| 29 | — | run_command (✓ ✓) |
| 30 | — | run_command (✓ ✓) |
Test case: task-07-generate-template-main
Target: agy/gemini-3.8-flash/high
crypto_rand prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-07-generate-template-main |
| Target | agy/gemini-3.8-flash/high |
| Prompt variant | crypto_rand |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/scripts/generate_template_main.txtar |
Vanilla LLM Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer from crypto/rand, ensuring all necessary standard library packages are imported cleanly. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer from crypto/rand, ensuring all necessary standard library packages are imported cleanly. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package main
func main() {
}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 74.92s | 141.31s | +88.6% | 124.24s | 235.30s | +89.4% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 19 | 26 | +36.8% | 30 | 54 | +80.0% |
| Initial Load / Discovery Turns | 10 | 5 | -50.0% | 19 | 8 | -57.9% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 18 | 25 | +38.9% | 29 | 53 | +82.8% |
| Output Tokens | 4805 | 8575 | +78.5% | 7511 | 12213 | +62.6% |
| Reasoning / Thinking Tokens | 2595 | 5807 | +123.8% | 3766 | 7023 | +86.5% |
| Total Input Tokens | 168657 | 189361 | +12.3% | 252660 | 390997 | +54.8% |
| Cached Input Tokens | 272148 | 511659 | +88.0% | 727622 | 1626229 | +123.5% |
| Uncached Input Tokens | 168657 | 189361 | +12.3% | 252660 | 390997 | +54.8% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 1.61:1 | 2.70:1 | +67.5% | 2.88:1 | 4.16:1 | +44.4% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 17 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 19 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | view_file (✓ ✓) |
| 3 | run_command (✓ ✓) | run_command (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | run_command (✓ ✓) | run_command (✓ ✓) |
| 6 | run_command (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 7 | run_command (✓ ✓) | view_file (✓ ✓) |
| 8 | run_command (✓ ✓) | run_command (✓ ✓) |
| 9 | view_file (✓ ✓) | run_command (✓ ✓) |
| 10 | run_command (✓ ✓) | run_command (✓ ✓) |
| 11 | write_to_file (✓ ✓) | view_file (✓ ✓) |
| 12 | run_command (✓ ✓) | view_file (✓ ✓) |
| 13 | run_command (✓ ✓) | run_command (✓ ✓) |
| 14 | run_command (✓ ✓) | view_file (✓ ✓) |
| 15 | run_command (✓ ✓) | run_command (✓ ✓) |
| 16 | run_command (✓ ✓) | run_command (✓ ✓) |
| 17 | view_file (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 18 | run_command (✓ ✓) | semedit/semantic_organize_imports (✓ ✓) |
| 19 | — | view_file (✓ ✓) |
| 20 | — | semedit/semantic_verify (✓ ✓) |
| 21 | — | run_command (✓ ✓) |
| 22 | — | run_command (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
| 24 | — | run_command (✓ ✓) |
| 25 | — | run_command (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 20 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 17 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | view_file (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | view_file (✓ ✓) |
| 4 | run_command (✓ ✓) | run_command (✓ ✓) |
| 5 | run_command (✓ ✓) | run_command (✓ ✓) |
| 6 | run_command (✓ ✓) | view_file (✓ ✓) |
| 7 | view_file (✓ ✓) | view_file (✓ ✓) |
| 8 | run_command (✓ ✓) | view_file (✓ ✓) |
| 9 | run_command (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 10 | view_file (✓ ✓) | run_command (✓ ✓) |
| 11 | run_command (✓ ✓) | run_command (✓ ✓) |
| 12 | run_command (✓ ✓) | run_command (✓ ✓) |
| 13 | run_command (✓ ✓) | run_command (✓ ?) |
| 14 | run_command (✓ ✓) | manage_task (✓ ✓) |
| 15 | run_command (✓ ✓) | view_file (✓ ✓) |
| 16 | run_command (✓ ✓) | view_file (✓ ✓) |
| 17 | view_file (✓ ✓) | run_command (✓ ✓) |
| 18 | view_file (✓ ✓) | run_command (✓ ✓) |
| 19 | run_command (✓ ✓) | run_command (✓ ✓) |
| 20 | replace_file_content (✓ ✓) | run_command (✓ ✓) |
| 21 | view_file (✓ ✓) | view_file (✓ ✓) |
| 22 | run_command (✓ ✓) | run_command (✓ ✓) |
| 23 | run_command (✓ ✓) | run_command (✓ ✓) |
| 24 | run_command (✓ ✓) | view_file (✓ ✓) |
| 25 | run_command (✓ ✓) | run_command (✓ ✓) |
| 26 | run_command (✓ ✓) | run_command (✓ ✓) |
| 27 | run_command (✓ ✓) | run_command (✓ ✓) |
| 28 | view_file (✓ ✓) | view_file (✓ ✓) |
| 29 | run_command (✓ ✓) | run_command (✓ ✓) |
| 30 | — | run_command (✓ ✓) |
| 31 | — | view_file (✓ ✓) |
| 32 | — | run_command (✓ ✓) |
| 33 | — | view_file (✓ ✓) |
| 34 | — | run_command (✓ ✓) |
| 35 | — | view_file (✓ ✓) |
| 36 | — | view_file (✓ ✓) |
| 37 | — | view_file (✓ ✓) |
| 38 | — | run_command (✓ ✓) |
| 39 | — | run_command (✓ ✓) |
| 40 | — | view_file (✓ ✓) |
| 41 | — | view_file (✓ ✓) |
| 42 | — | run_command (✓ ✓) |
| 43 | — | view_file (✓ ✓) |
| 44 | — | view_file (✓ ✓) |
| 45 | — | view_file (✓ ✓) |
| 46 | — | run_command (✓ ✓) |
| 47 | — | view_file (✓ ✓) |
| 48 | — | semedit/semantic_batch (✓ ✓) |
| 49 | — | view_file (✓ ✓) |
| 50 | — | run_command (✓ ✓) |
| 51 | — | run_command (✓ ✓) |
| 52 | — | run_command (✓ ✓) |
| 53 | — | semedit/semantic_verify (✓ ✓) |
default prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-07-generate-template-main |
| Target | agy/gemini-3.8-flash/high |
| Prompt variant | default |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/scripts/generate_template_main.txtar |
Vanilla LLM Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer, ensuring all necessary standard library packages are imported cleanly. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer, ensuring all necessary standard library packages are imported cleanly. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package main
func main() {
}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 48.76s | 119.05s | +144.2% | 63.20s | 142.06s | +124.8% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 14 | 25 | +78.6% | 13 | 30 | +130.8% |
| Initial Load / Discovery Turns | 8 | 12 | +50.0% | 6 | 5 | -16.7% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 13 | 24 | +84.6% | 12 | 29 | +141.7% |
| Output Tokens | 3515 | 7116 | +102.4% | 2980 | 8753 | +193.7% |
| Reasoning / Thinking Tokens | 1913 | 4640 | +142.6% | 1556 | 5590 | +259.3% |
| Total Input Tokens | 115989 | 220064 | +89.7% | 104261 | 202705 | +94.4% |
| Cached Input Tokens | 207357 | 495769 | +139.1% | 182950 | 935971 | +411.6% |
| Uncached Input Tokens | 115989 | 220064 | +89.7% | 104261 | 202705 | +94.4% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 1.79:1 | 2.25:1 | +26.0% | 1.75:1 | 4.62:1 | +163.1% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 12 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 12 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | view_file (✓ ✓) |
| 2 | view_file (✓ ✓) | run_command (✓ ✓) |
| 3 | view_file (✓ ✓) | view_file (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | run_command (✓ ✓) | run_command (✓ ✓) |
| 6 | run_command (✓ ✓) | run_command (✓ ✓) |
| 7 | view_file (✓ ✓) | run_command (✓ ✓) |
| 8 | run_command (✓ ✓) | run_command (✓ ✓) |
| 9 | write_to_file (✓ ✓) | run_command (✓ ✓) |
| 10 | run_command (✓ ✓) | view_file (✓ ✓) |
| 11 | run_command (✓ ✓) | view_file (✓ ✓) |
| 12 | run_command (✓ ✓) | view_file (✓ ✓) |
| 13 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 14 | — | run_command (✓ ✓) |
| 15 | — | view_file (✓ ✓) |
| 16 | — | run_command (✓ ✓) |
| 17 | — | view_file (✓ ✓) |
| 18 | — | semedit/semantic_replace_body (✓ ✓) |
| 19 | — | view_file (✓ ✓) |
| 20 | — | semedit/semantic_verify (✓ ✓) |
| 21 | — | run_command (✓ ✓) |
| 22 | — | run_command (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
| 24 | — | view_file (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 12 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 12 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | view_file (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | view_file (✓ ✓) |
| 4 | view_file (✓ ✓) | run_command (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 7 | write_to_file (✓ ✓) | view_file (✓ ✓) |
| 8 | run_command (✓ ✓) | run_command (✓ ✓) |
| 9 | run_command (✓ ✓) | run_command (✓ ✓) |
| 10 | run_command (✓ ✓) | view_file (✓ ✓) |
| 11 | run_command (✓ ✓) | view_file (✓ ✓) |
| 12 | view_file (✓ ✓) | view_file (✓ ✓) |
| 13 | — | run_command (✓ ✓) |
| 14 | — | view_file (✓ ✓) |
| 15 | — | view_file (✓ ✓) |
| 16 | — | view_file (✓ ✓) |
| 17 | — | run_command (✓ ✓) |
| 18 | — | run_command (✓ ✓) |
| 19 | — | semedit/semantic_replace_body (✓ ✓) |
| 20 | — | semedit/semantic_organize_imports (✓ ✓) |
| 21 | — | view_file (✓ ✓) |
| 22 | — | run_command (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
| 24 | — | run_command (✓ ✓) |
| 25 | — | run_command (✓ ✓) |
| 26 | — | run_command (✓ ✓) |
| 27 | — | view_file (✓ ✓) |
| 28 | — | semedit/semantic_verify (✓ ✓) |
| 29 | — | run_command (✓ ✓) |
prefer_discover_semedit prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-07-generate-template-main |
| Target | agy/gemini-3.8-flash/high |
| Prompt variant | prefer_discover_semedit |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/scripts/generate_template_main.txtar |
Vanilla LLM Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
Before editing, inspect the complete available tool inventory, including deferred or lazy tools. If applicable semantic editing tools are callable, prefer them for source mutations. In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer, ensuring all necessary standard library packages are imported cleanly. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
Before editing, inspect the complete available tool inventory, including deferred or lazy tools. If applicable semantic editing tools are callable, prefer them for source mutations. In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer, ensuring all necessary standard library packages are imported cleanly. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package main
func main() {
}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 71.46s | 171.49s | +140.0% | 84.04s | 140.82s | +67.6% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 13 | 38 | +192.3% | 19 | 29 | +52.6% |
| Initial Load / Discovery Turns | 6 | 10 | +66.7% | 11 | 6 | -45.5% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 12 | 37 | +208.3% | 18 | 28 | +55.6% |
| Output Tokens | 6077 | 11096 | +82.6% | 5827 | 6699 | +15.0% |
| Reasoning / Thinking Tokens | 4557 | 6904 | +51.5% | 3633 | 3719 | +2.4% |
| Total Input Tokens | 128096 | 355146 | +177.2% | 100029 | 246828 | +146.8% |
| Cached Input Tokens | 183040 | 1028088 | +461.7% | 414785 | 682914 | +64.6% |
| Uncached Input Tokens | 128096 | 355146 | +177.2% | 100029 | 246828 | +146.8% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 1.43:1 | 2.89:1 | +102.6% | 4.15:1 | 2.77:1 | -33.3% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 12 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 12 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | view_file (✓ ✓) | run_command (✓ ✓) |
| 3 | view_file (✓ ✓) | view_file (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | view_file (✓ ✓) |
| 7 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 8 | run_command (✓ ✓) | run_command (✓ ✓) |
| 9 | run_command (✓ ✓) | run_command (✓ ✓) |
| 10 | run_command (✓ ✓) | view_file (✓ ✓) |
| 11 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 12 | run_command (✓ ✓) | run_command (✓ ✓) |
| 13 | — | run_command (✓ ✓) |
| 14 | — | run_command (✓ ✓) |
| 15 | — | run_command (✓ ✓) |
| 16 | — | run_command (✓ ✓) |
| 17 | — | run_command (✓ ✓) |
| 18 | — | run_command (✓ ✓) |
| 19 | — | run_command (✓ ✓) |
| 20 | — | run_command (✓ ✓) |
| 21 | — | view_file (✓ ✓) |
| 22 | — | run_command (✓ ✓) |
| 23 | — | view_file (✓ ✓) |
| 24 | — | run_command (✓ ✓) |
| 25 | — | view_file (✓ ✓) |
| 26 | — | view_file (✓ ✓) |
| 27 | — | view_file (✓ ✓) |
| 28 | — | semedit/semantic_replace_body (✓ ✓) |
| 29 | — | semedit/semantic_organize_imports (✓ ✓) |
| 30 | — | view_file (✓ ✓) |
| 31 | — | run_command (✓ ✓) |
| 32 | — | run_command (✓ ✓) |
| 33 | — | run_command (✓ ✓) |
| 34 | — | run_command (✓ ✓) |
| 35 | — | run_command (✓ ✓) |
| 36 | — | view_file (✓ ✓) |
| 37 | — | semedit/semantic_verify (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 12 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 12 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | view_file (✓ ✓) | view_file (✓ ✓) |
| 3 | view_file (✓ ✓) | run_command (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | run_command (✓ ✓) | view_file (✓ ✓) |
| 7 | run_command (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 8 | run_command (✓ ✓) | view_file (✓ ✓) |
| 9 | run_command (✓ ✓) | run_command (✓ ✓) |
| 10 | run_command (✓ ✓) | run_command (✓ ✓) |
| 11 | view_file (✓ ✓) | run_command (✓ ✓) |
| 12 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 13 | run_command (✓ ✓) | run_command (✓ ✓) |
| 14 | run_command (✓ ✓) | view_file (✓ ✓) |
| 15 | run_command (✓ ✓) | run_command (✓ ✓) |
| 16 | run_command (✓ ✓) | view_file (✓ ✓) |
| 17 | run_command (✓ ✓) | run_command (✓ ✓) |
| 18 | view_file (✓ ✓) | view_file (✓ ✓) |
| 19 | — | view_file (✓ ✓) |
| 20 | — | semedit/semantic_replace_body (✓ ✓) |
| 21 | — | semedit/semantic_organize_imports (✓ ✓) |
| 22 | — | view_file (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
| 24 | — | run_command (✓ ✓) |
| 25 | — | run_command (✓ ✓) |
| 26 | — | view_file (✓ ✓) |
| 27 | — | run_command (✓ ✓) |
| 28 | — | view_file (✓ ✓) |
Target: agy/gemini-3.8-flash/medium
crypto_rand prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-07-generate-template-main |
| Target | agy/gemini-3.8-flash/medium |
| Prompt variant | crypto_rand |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/scripts/generate_template_main.txtar |
Vanilla LLM Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer from crypto/rand, ensuring all necessary standard library packages are imported cleanly. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer from crypto/rand, ensuring all necessary standard library packages are imported cleanly. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package main
func main() {
}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 76.88s | 73.55s | -4.3% | 68.32s | 208.87s | +205.7% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 16 | 18 | +12.5% | 19 | 39 | +105.3% |
| Initial Load / Discovery Turns | 10 | 12 | +20.0% | 11 | 21 | +90.9% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 15 | 17 | +13.3% | 18 | 38 | +111.1% |
| Output Tokens | 3991 | 4713 | +18.1% | 5373 | 10023 | +86.5% |
| Reasoning / Thinking Tokens | 2254 | 2801 | +24.3% | 3183 | 5348 | +68.0% |
| Total Input Tokens | 150345 | 98320 | -34.6% | 138779 | 354164 | +155.2% |
| Cached Input Tokens | 349437 | 337293 | -3.5% | 337311 | 1105369 | +227.7% |
| Uncached Input Tokens | 150345 | 98320 | -34.6% | 138779 | 354164 | +155.2% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 2.32:1 | 3.43:1 | +47.6% | 2.43:1 | 3.12:1 | +28.4% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 20 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 21 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | view_file (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | view_file (✓ ✓) |
| 4 | view_file (✓ ✓) | view_file (✓ ✗): Created At: 2026-09-24T18:24:50+02:00 Completed At: 2026-09-24T18:24:50+02:00 Encountered error in step execution: declaring permissions: cortex tool view_file: convert tool call for permissions: model output error: invalid tool call error (invalid_args) failed to read file: stat /Users/alessandro/.gemini/antigravity-cli/mcp/semedit/instructions.md: no such file or directory |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | run_command (✓ ✓) |
| 7 | view_file (✓ ✓) | view_file (✓ ✓) |
| 8 | run_command (✓ ✓) | run_command (✓ ✓) |
| 9 | run_command (✓ ✓) | view_file (✓ ✓) |
| 10 | run_command (✓ ✓) | run_command (✓ ✓) |
| 11 | write_to_file (✓ ✓) | view_file (✓ ✓) |
| 12 | run_command (✓ ✓) | view_file (✓ ✓) |
| 13 | run_command (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 14 | run_command (✓ ✓) | view_file (✓ ✓) |
| 15 | view_file (✓ ✓) | run_command (✓ ✓) |
| 16 | — | run_command (✓ ✓) |
| 17 | — | run_command (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 14 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 19 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | view_file (✓ ✓) |
| 3 | run_command (✓ ✓) | view_file (✓ ✗): Created At: 2026-09-24T18:30:06+02:00 Completed At: 2026-09-24T18:30:06+02:00 Encountered error in step execution: declaring permissions: cortex tool view_file: convert tool call for permissions: model output error: invalid tool call error (invalid_args) failed to read file: stat /Users/alessandro/.gemini/antigravity-cli/mcp/semedit/instructions.md: no such file or directory |
| 4 | run_command (✓ ✓) | run_command (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | view_file (✓ ✓) |
| 7 | view_file (✓ ✓) | run_command (✓ ✓) |
| 8 | run_command (✓ ✓) | view_file (✓ ✓) |
| 9 | view_file (✓ ✓) | view_file (✓ ✓) |
| 10 | run_command (✓ ✓) | run_command (✓ ✓) |
| 11 | run_command (✓ ✓) | run_command (✓ ✓) |
| 12 | write_to_file (✓ ✓) | run_command (✓ ✓) |
| 13 | run_command (✓ ✓) | run_command (✓ ✓) |
| 14 | run_command (✓ ✓) | run_command (✓ ✓) |
| 15 | run_command (✓ ✓) | run_command (✓ ✓) |
| 16 | run_command (✓ ✓) | run_command (✓ ✓) |
| 17 | run_command (✓ ✓) | run_command (✓ ✓) |
| 18 | run_command (✓ ✓) | run_command (✓ ✓) |
| 19 | — | view_file (✓ ✓) |
| 20 | — | view_file (✓ ✓) |
| 21 | — | view_file (✓ ✓) |
| 22 | — | semedit/semantic_organize_imports (✓ ✓) |
| 23 | — | view_file (✓ ✓) |
| 24 | — | run_command (✓ ✓) |
| 25 | — | semedit/semantic_replace_body (✓ ✓) |
| 26 | — | semedit/semantic_organize_imports (✓ ✓) |
| 27 | — | view_file (✓ ✓) |
| 28 | — | semedit/semantic_verify (✓ ✓) |
| 29 | — | run_command (✓ ✓) |
| 30 | — | run_command (✓ ✓) |
| 31 | — | run_command (✓ ✓) |
| 32 | — | view_file (✓ ✓) |
| 33 | — | run_command (✓ ✓) |
| 34 | — | run_command (✓ ✓) |
| 35 | — | run_command (✓ ✓) |
| 36 | — | run_command (✓ ✓) |
| 37 | — | run_command (✓ ✓) |
| 38 | — | run_command (✓ ✓) |
default prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-07-generate-template-main |
| Target | agy/gemini-3.8-flash/medium |
| Prompt variant | default |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/scripts/generate_template_main.txtar |
Vanilla LLM Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer, ensuring all necessary standard library packages are imported cleanly. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer, ensuring all necessary standard library packages are imported cleanly. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package main
func main() {
}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 39.85s | 123.36s | +209.5% | 87.73s | 109.60s | +24.9% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 11 | 27 | +145.5% | 16 | 17 | +6.2% |
| Initial Load / Discovery Turns | 6 | 5 | -16.7% | 10 | 4 | -60.0% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 10 | 26 | +160.0% | 15 | 16 | +6.7% |
| Output Tokens | 2528 | 7054 | +179.0% | 3368 | 3574 | +6.1% |
| Reasoning / Thinking Tokens | 1324 | 4488 | +239.0% | 1527 | 1773 | +16.1% |
| Total Input Tokens | 185215 | 232904 | +25.7% | 117558 | 155571 | +32.3% |
| Cached Input Tokens | 32514 | 536202 | +1549.1% | 243796 | 198913 | -18.4% |
| Uncached Input Tokens | 152701 | 232904 | +52.5% | 117558 | 155571 | +32.3% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 0.21:1 | 2.30:1 | +981.2% | 2.07:1 | 1.28:1 | -38.3% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 12 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 12 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | view_file (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | view_file (✓ ✓) |
| 4 | view_file (✓ ✓) | view_file (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 7 | write_to_file (✓ ✓) | run_command (✓ ✓) |
| 8 | run_command (✓ ✓) | run_command (✓ ✓) |
| 9 | run_command (✓ ✓) | run_command (✓ ✓) |
| 10 | run_command (✓ ✓) | view_file (✓ ✓) |
| 11 | — | run_command (✓ ✓) |
| 12 | — | run_command (✓ ✓) |
| 13 | — | run_command (✓ ✓) |
| 14 | — | run_command (✓ ✓) |
| 15 | — | run_command (✓ ✓) |
| 16 | — | run_command (✓ ✓) |
| 17 | — | run_command (✓ ✓) |
| 18 | — | view_file (✓ ✓) |
| 19 | — | view_file (✓ ✓) |
| 20 | — | semedit/semantic_batch (✓ ✓) |
| 21 | — | view_file (✓ ✓) |
| 22 | — | run_command (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
| 24 | — | run_command (✓ ✓) |
| 25 | — | run_command (✓ ✓) |
| 26 | — | view_file (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 12 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 14 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | view_file (✓ ✓) |
| 2 | view_file (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | view_file (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | view_file (✓ ✓) | workspace-servers_semedit/semantic_replace_body (✓ ✗): Created At: 2026-09-24T18:31:01+02:00 Completed At: 2026-09-24T18:31:01+02:00 Encountered error in step execution: read target file: open /Users/alessandro/sources/semantic-editor/.agents/plugins/workspace-servers/main.go: no such file or directory |
| 6 | run_command (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 7 | run_command (✓ ✓) | view_file (✓ ✓) |
| 8 | run_command (✓ ✓) | view_file (✓ ✓) |
| 9 | run_command (✓ ✓) | semedit/semantic_organize_imports (✓ ✓) |
| 10 | view_file (✓ ✓) | view_file (✓ ✓) |
| 11 | replace_file_content (✓ ✓) | run_command (✓ ✓) |
| 12 | view_file (✓ ✓) | run_command (✓ ✓) |
| 13 | run_command (✓ ✓) | run_command (✓ ✓) |
| 14 | run_command (✓ ✓) | run_command (✓ ✓) |
| 15 | run_command (✓ ✓) | run_command (✓ ✓) |
| 16 | — | run_command (✓ ✓) |
prefer_discover_semedit prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-07-generate-template-main |
| Target | agy/gemini-3.8-flash/medium |
| Prompt variant | prefer_discover_semedit |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/scripts/generate_template_main.txtar |
Vanilla LLM Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
Before editing, inspect the complete available tool inventory, including deferred or lazy tools. If applicable semantic editing tools are callable, prefer them for source mutations. In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer, ensuring all necessary standard library packages are imported cleanly. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
You are forbidden to modify protected files: “go.mod”, “go.sum”.
Before editing, inspect the complete available tool inventory, including deferred or lazy tools. If applicable semantic editing tools are callable, prefer them for source mutations. In main.go, implement main() to execute a text/template that prints ‘Hello World’ along with a random integer, ensuring all necessary standard library packages are imported cleanly. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package main
func main() {
}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 80.47s | 73.05s | -9.2% | 45.54s | 96.69s | +112.3% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 20 | 18 | -10.0% | 9 | 24 | +166.7% |
| Initial Load / Discovery Turns | 8 | 6 | -25.0% | 3 | 12 | +300.0% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 19 | 17 | -10.5% | 8 | 23 | +187.5% |
| Output Tokens | 5094 | 4111 | -19.3% | 2839 | 5339 | +88.1% |
| Reasoning / Thinking Tokens | 2795 | 2081 | -25.5% | 1782 | 3014 | +69.1% |
| Total Input Tokens | 231406 | 173139 | -25.2% | 60475 | 155576 | +157.3% |
| Cached Input Tokens | 268213 | 211076 | -21.3% | 113815 | 475194 | +317.5% |
| Uncached Input Tokens | 231406 | 173139 | -25.2% | 60475 | 155576 | +157.3% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 1.16:1 | 1.22:1 | +5.2% | 1.88:1 | 3.05:1 | +62.3% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 12 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File main.go modified (4 lines -> 14 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | view_file (✓ ✓) | run_command (✓ ✓) |
| 3 | view_file (✓ ✓) | view_file (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | run_command (✓ ✓) | view_file (✓ ✓) |
| 7 | view_file (✓ ✓) | workspace-servers_semedit/semantic_replace_body (✓ ✗): Created At: 2026-09-24T18:28:10+02:00 Completed At: 2026-09-24T18:28:13+02:00 Encountered error in step execution: read target file: open /Users/alessandro/sources/semantic-editor/.agents/plugins/workspace-servers/main.go: no such file or directory |
| 8 | view_file (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 9 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 10 | run_command (✓ ?) | semedit/semantic_organize_imports (✓ ✓) |
| 11 | manage_task (✓ ✓) | view_file (✓ ✓) |
| 12 | run_command (✓ ✓) | run_command (✓ ✓) |
| 13 | run_command (✓ ✓) | run_command (✓ ✓) |
| 14 | run_command (✓ ✓) | run_command (✓ ✓) |
| 15 | view_file (✓ ✓) | run_command (✓ ✓) |
| 16 | run_command (✓ ✓) | run_command (✓ ✓) |
| 17 | run_command (✓ ✓) | view_file (✓ ✓) |
| 18 | view_file (✓ ✓) | — |
| 19 | run_command (✓ ✓) | — |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 15 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File main.go modified (4 lines -> 12 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | view_file (✓ ✓) |
| 2 | run_command (✓ ✓) | view_file (✓ ✓) |
| 3 | view_file (✓ ✓) | view_file (✓ ✓) |
| 4 | write_to_file (✓ ✓) | run_command (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✗): Created At: 2026-09-24T18:32:57+02:00 Completed At: 2026-09-24T18:32:57+02:00 Encountered error in step execution: declaring permissions: cortex tool view_file: convert tool call for permissions: model output error: invalid tool call error (invalid_args) failed to read file: stat /Users/alessandro/.gemini/antigravity-cli/mcp/semedit/instructions.md: no such file or directory |
| 6 | run_command (✓ ✓) | view_file (✓ ✓) |
| 7 | run_command (✓ ✓) | run_command (✓ ✓) |
| 8 | view_file (✓ ✓) | run_command (✓ ✓) |
| 9 | — | run_command (✓ ✓) |
| 10 | — | run_command (✓ ✓) |
| 11 | — | run_command (✓ ✓) |
| 12 | — | view_file (✓ ✓) |
| 13 | — | workspace-servers_semedit/semantic_organize_imports (✓ ✗): Created At: 2026-09-24T18:33:35+02:00 Completed At: 2026-09-24T18:33:35+02:00 Encountered error in step execution: stat path /Users/alessandro/sources/semantic-editor/.agents/plugins/workspace-servers/main.go: stat /Users/alessandro/sources/semantic-editor/.agents/plugins/workspace-servers/main.go: no such file or directory |
| 14 | — | semedit/semantic_organize_imports (✓ ✓) |
| 15 | — | view_file (✓ ✓) |
| 16 | — | semedit/semantic_replace_body (✓ ✓) |
| 17 | — | semedit/semantic_organize_imports (✓ ✓) |
| 18 | — | view_file (✓ ✓) |
| 19 | — | semedit/semantic_verify (✓ ✓) |
| 20 | — | run_command (✓ ✓) |
| 21 | — | run_command (✓ ✓) |
| 22 | — | run_command (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
Test case: task-09-composite-refactor
Target: agy/gemini-3.8-flash/high
default prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-09-composite-refactor |
| Target | agy/gemini-3.8-flash/high |
| Prompt variant | default |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/bench/task_09_composite_refactor.txtar |
Vanilla LLM Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
In config/config.go: rename Config.Port to Config.ListenPort, insert helper function DefaultConfig() *Config returning default values, and clean up unused imports. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
In config/config.go: rename Config.Port to Config.ListenPort, insert helper function DefaultConfig() *Config returning default values, and clean up unused imports. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package config
import (
"archive/tar"
"fmt"
)
type Config struct {
Host string
Port int
}
func (c *Config) Addr() string {
return fmt.Sprintf("%s:%d", c.Host, c.Port)
}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 82.66s | 105.67s | +27.8% | 58.14s | 140.84s | +142.2% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 24 | 24 | 0% | 16 | 40 | +150.0% |
| Initial Load / Discovery Turns | 11 | 13 | +18.2% | 10 | 13 | +30.0% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 23 | 23 | 0% | 15 | 39 | +160.0% |
| Output Tokens | 6623 | 8527 | +28.7% | 4917 | 9887 | +101.1% |
| Reasoning / Thinking Tokens | 3776 | 6211 | +64.5% | 3113 | 6013 | +93.2% |
| Total Input Tokens | 179412 | 184124 | +2.6% | 152182 | 340917 | +124.0% |
| Cached Input Tokens | 483508 | 471303 | -2.5% | 219458 | 904531 | +312.2% |
| Uncached Input Tokens | 179412 | 184124 | +2.6% | 152182 | 340917 | +124.0% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 2.69:1 | 2.56:1 | -5.0% | 1.44:1 | 2.65:1 | +84.0% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File config/config.go modified (15 lines -> 21 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File config/config.go modified (15 lines -> 21 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | view_file (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | view_file (✓ ✓) | run_command (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | run_command (✓ ✓) | run_command (✓ ✓) |
| 6 | run_command (✓ ?) | view_file (✓ ✓) |
| 7 | manage_task (✓ ✓) | run_command (✓ ✓) |
| 8 | manage_task (✓ ✗): Created At: 2026-09-24T18:54:55+02:00 Completed At: 2026-09-24T18:54:55+02:00 Encountered error in step execution: cannot kill task "ef24fa45-818d-4f5c-b740-8f042b62e7c1/task-12": task is not running (status: DONE) | view_file (✓ ✓) |
| 9 | run_command (✓ ✓) | view_file (✓ ✓) |
| 10 | run_command (✓ ✓) | view_file (✓ ✓) |
| 11 | run_command (✓ ✓) | view_file (✓ ✓) |
| 12 | replace_file_content (✓ ✓) | run_command (✓ ✓) |
| 13 | view_file (✓ ✓) | view_file (✓ ✓) |
| 14 | run_command (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 15 | run_command (✓ ✓) | run_command (✓ ✓) |
| 16 | run_command (✓ ✓) | semedit/semantic_batch (✓ ✓) |
| 17 | run_command (✓ ✓) | view_file (✓ ✓) |
| 18 | run_command (✓ ✓) | run_command (✓ ✓) |
| 19 | run_command (✓ ✓) | run_command (✓ ✓) |
| 20 | view_file (✓ ✓) | view_file (✓ ✓) |
| 21 | run_command (✓ ✓) | view_file (✓ ✓) |
| 22 | run_command (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 23 | view_file (✓ ✓) | run_command (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File config/config.go modified (15 lines -> 21 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File config/config.go modified (15 lines -> 21 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | view_file (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | run_command (✓ ✓) |
| 4 | view_file (✓ ✓) | run_command (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | run_command (✓ ✓) |
| 7 | view_file (✓ ✓) | run_command (✓ ✓) |
| 8 | run_command (✓ ✓) | run_command (✓ ✓) |
| 9 | run_command (✓ ✓) | view_file (✓ ✓) |
| 10 | run_command (✓ ✓) | view_file (✓ ✓) |
| 11 | write_to_file (✓ ✓) | view_file (✓ ✓) |
| 12 | run_command (✓ ✓) | view_file (✓ ✓) |
| 13 | run_command (✓ ✓) | view_file (✓ ✓) |
| 14 | run_command (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 15 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 16 | — | run_command (✓ ✓) |
| 17 | — | run_command (✓ ✓) |
| 18 | — | view_file (✓ ✓) |
| 19 | — | run_command (✓ ✓) |
| 20 | — | semedit/semantic_rename (✓ ✓) |
| 21 | — | view_file (✓ ✓) |
| 22 | — | semedit/semantic_organize_imports (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
| 24 | — | view_file (✓ ✓) |
| 25 | — | semedit/semantic_insert_function (✓ ✓) |
| 26 | — | view_file (✓ ✓) |
| 27 | — | semedit/semantic_organize_imports (✓ ✓) |
| 28 | — | semedit/semantic_verify (✓ ✓) |
| 29 | — | run_command (✓ ✓) |
| 30 | — | run_command (✓ ✓) |
| 31 | — | run_command (✓ ✓) |
| 32 | — | run_command (✓ ✓) |
| 33 | — | run_command (✓ ✓) |
| 34 | — | run_command (✓ ✓) |
| 35 | — | view_file (✓ ✓) |
| 36 | — | view_file (✓ ✓) |
| 37 | — | run_command (✓ ✓) |
| 38 | — | run_command (✓ ✓) |
| 39 | — | semedit/semantic_verify (✓ ✓) |
Target: agy/gemini-3.8-flash/medium
default prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-09-composite-refactor |
| Target | agy/gemini-3.8-flash/medium |
| Prompt variant | default |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/bench/task_09_composite_refactor.txtar |
Vanilla LLM Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
In config/config.go: rename Config.Port to Config.ListenPort, insert helper function DefaultConfig() *Config returning default values, and clean up unused imports. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
In config/config.go: rename Config.Port to Config.ListenPort, insert helper function DefaultConfig() *Config returning default values, and clean up unused imports. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
package config
import (
"archive/tar"
"fmt"
)
type Config struct {
Host string
Port int
}
func (c *Config) Addr() string {
return fmt.Sprintf("%s:%d", c.Host, c.Port)
}| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 62.35s | 70.25s | +12.7% | 85.96s | 82.23s | -4.3% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 15 | 19 | +26.7% | 23 | 24 | +4.3% |
| Initial Load / Discovery Turns | 8 | 5 | -37.5% | 14 | 10 | -28.6% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 14 | 18 | +28.6% | 22 | 23 | +4.5% |
| Output Tokens | 3848 | 4713 | +22.5% | 5170 | 6179 | +19.5% |
| Reasoning / Thinking Tokens | 2088 | 2819 | +35.0% | 2724 | 3908 | +43.5% |
| Total Input Tokens | 115558 | 179338 | +55.2% | 238253 | 222158 | -6.8% |
| Cached Input Tokens | 248053 | 325276 | +31.1% | 418730 | 438859 | +4.8% |
| Uncached Input Tokens | 115558 | 179338 | +55.2% | 238253 | 222158 | -6.8% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 2.15:1 | 1.81:1 | -15.5% | 1.76:1 | 1.98:1 | +12.4% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File config/config.go modified (15 lines -> 21 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File config/config.go modified (15 lines -> 21 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | view_file (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | view_file (✓ ✓) |
| 4 | view_file (✓ ✓) | run_command (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | run_command (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 7 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 8 | view_file (✓ ✓) | view_file (✓ ✓) |
| 9 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 10 | run_command (✓ ✓) | semedit/semantic_batch (✓ ✓) |
| 11 | run_command (✓ ✓) | view_file (✓ ✓) |
| 12 | run_command (✓ ✓) | run_command (✓ ✓) |
| 13 | view_file (✓ ✓) | run_command (✓ ✓) |
| 14 | run_command (✓ ✓) | run_command (✓ ✓) |
| 15 | — | run_command (✓ ✓) |
| 16 | — | run_command (✓ ✓) |
| 17 | — | view_file (✓ ✓) |
| 18 | — | run_command (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File config/config.go modified (15 lines -> 21 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File config/config.go modified (15 lines -> 21 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | view_file (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | view_file (✓ ✓) |
| 3 | run_command (✓ ✓) | view_file (✓ ✓) |
| 4 | run_command (✓ ✓) | run_command (✓ ✓) |
| 5 | view_file (✓ ✓) | run_command (✓ ✓) |
| 6 | run_command (✓ ✓) | run_command (✓ ✓) |
| 7 | run_command (✓ ✓) | view_file (✓ ✓) |
| 8 | run_command (✓ ✓) | view_file (✓ ✓) |
| 9 | run_command (✓ ?) | view_file (✓ ✓) |
| 10 | manage_task (✓ ✓) | view_file (✓ ✓) |
| 11 | manage_task (✓ ✗): Created At: 2026-09-24T18:53:39+02:00 Completed At: 2026-09-24T18:53:39+02:00 Encountered error in step execution: cannot kill task "2aa7db55-556b-492d-b60e-9ea7d55d08e5/task-18": task is not running (status: DONE) | semedit/semantic_rename (✓ ✓) |
| 12 | run_command (✓ ✓) | view_file (✓ ✓) |
| 13 | run_command (✓ ✓) | run_command (✓ ✓) |
| 14 | view_file (✓ ✓) | view_file (✓ ✓) |
| 15 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 16 | view_file (✓ ✓) | semedit/semantic_insert_function (✓ ✓) |
| 17 | run_command (✓ ✓) | view_file (✓ ✓) |
| 18 | run_command (✓ ✓) | semedit/semantic_organize_imports (✓ ✓) |
| 19 | run_command (✓ ✓) | run_command (✓ ✓) |
| 20 | run_command (✓ ✓) | run_command (✓ ✓) |
| 21 | view_file (✓ ✓) | run_command (✓ ✓) |
| 22 | run_command (✓ ✓) | run_command (✓ ✓) |
| 23 | — | run_command (✓ ✓) |
Test case: task-11-mixed-sink-api-migration
Target: agy/gemini-3.8-flash/high
default prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-11-mixed-sink-api-migration |
| Target | agy/gemini-3.8-flash/high |
| Prompt variant | default |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/bench/task_11_mixed_sink_api_migration.txtar |
Vanilla LLM Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Audit event kinds may contain inconsistent whitespace and casing. Normalize them at the delivery boundary before persistence, and treat probe events as control traffic. Do not change the metrics or legacy export protocols. Preserve the module metadata and tests, and leave the workspace verified. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Audit event kinds may contain inconsistent whitespace and casing. Normalize them at the delivery boundary before persistence, and treat probe events as control traffic. Do not change the metrics or legacy export protocols. Preserve the module metadata and tests, and leave the workspace verified. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
// Package audit leaves room for normalization rules owned by the delivery boundary.
package audit| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 289.35s | 175.97s | -39.2% | — | — | — |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | — | — | — |
| Internal Tool Cycles | 71 | 49 | -31.0% | — | — | — |
| Initial Load / Discovery Turns | 34 | 25 | -26.5% | — | — | — |
| MCP Discovery / Schema Turns | 0 | 0 | — | — | — | — |
| Tool Invocations (per run) | 70 | 48 | -31.4% | — | — | — |
| Output Tokens | 19658 | 13247 | -32.6% | — | — | — |
| Reasoning / Thinking Tokens | 10965 | 7866 | -28.3% | — | — | — |
| Total Input Tokens | 601319 | 402562 | -33.1% | — | — | — |
| Cached Input Tokens | 2787269 | 1395729 | -49.9% | — | — | — |
| Uncached Input Tokens | 601319 | 402562 | -33.1% | — | — | — |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 4.64:1 | 3.47:1 | -25.2% | — | — | — |
| Oracle L1: Mutation Policy | PASS | PASS | — | — | — | — |
| Oracle L2: AST Invariants | PASS | PASS | — | — | — | — |
| Oracle L3: Clean Build | PASS | PASS | — | — | — | — |
| Oracle L4: Verification Test | PASS | PASS | — | — | — | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | — | — | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | run_command (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | view_file (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | view_file (✓ ✓) |
| 7 | view_file (✓ ✓) | view_file (✓ ✓) |
| 8 | view_file (✓ ✓) | view_file (✓ ✓) |
| 9 | view_file (✓ ✓) | view_file (✓ ✓) |
| 10 | view_file (✓ ✓) | view_file (✓ ✓) |
| 11 | view_file (✓ ✓) | view_file (✓ ✓) |
| 12 | view_file (✓ ✓) | view_file (✓ ✓) |
| 13 | view_file (✓ ✓) | view_file (✓ ✓) |
| 14 | view_file (✓ ✓) | view_file (✓ ✓) |
| 15 | view_file (✓ ✓) | view_file (✓ ✓) |
| 16 | view_file (✓ ✓) | view_file (✓ ✓) |
| 17 | view_file (✓ ✓) | view_file (✓ ✓) |
| 18 | view_file (✓ ✓) | run_command (✓ ✓) |
| 19 | view_file (✓ ✓) | view_file (✓ ✓) |
| 20 | run_command (✓ ✓) | run_command (✓ ✓) |
| 21 | run_command (✓ ✓) | run_command (✓ ✓) |
| 22 | run_command (✓ ✓) | run_command (✓ ✓) |
| 23 | run_command (✓ ✓) | run_command (✓ ✓) |
| 24 | view_file (✓ ✓) | run_command (✓ ✓) |
| 25 | view_file (✓ ✓) | view_file (✓ ✓) |
| 26 | run_command (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 27 | run_command (✓ ✓) | view_file (✓ ✓) |
| 28 | view_file (✓ ✓) | view_file (✓ ✓) |
| 29 | view_file (✓ ✓) | view_file (✓ ✓) |
| 30 | run_command (✓ ✓) | semedit/semantic_insert_declaration (✓ ✓) |
| 31 | run_command (✓ ✓) | view_file (✓ ✓) |
| 32 | run_command (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 33 | run_command (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 34 | run_command (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 35 | replace_file_content (✓ ✓) | run_command (✓ ✓) |
| 36 | view_file (✓ ✓) | run_command (✓ ✓) |
| 37 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 38 | view_file (✓ ✓) | view_file (✓ ✓) |
| 39 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 40 | run_command (✓ ✓) | view_file (✓ ✓) |
| 41 | run_command (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 42 | run_command (✓ ✓) | run_command (✓ ✓) |
| 43 | run_command (✓ ✓) | run_command (✓ ✓) |
| 44 | run_command (✓ ✓) | run_command (✓ ✓) |
| 45 | run_command (✓ ✓) | run_command (✓ ✓) |
| 46 | run_command (✓ ✓) | run_command (✓ ✓) |
| 47 | view_file (✓ ✓) | run_command (✓ ✓) |
| 48 | run_command (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 49 | view_file (✓ ✓) | — |
| 50 | run_command (✓ ✓) | — |
| 51 | run_command (✓ ✓) | — |
| 52 | view_file (✓ ✓) | — |
| 53 | view_file (✓ ✓) | — |
| 54 | view_file (✓ ✓) | — |
| 55 | view_file (✓ ✓) | — |
| 56 | run_command (✓ ✓) | — |
| 57 | run_command (✓ ✓) | — |
| 58 | run_command (✓ ✓) | — |
| 59 | run_command (✓ ✓) | — |
| 60 | run_command (✓ ✓) | — |
| 61 | run_command (✓ ✓) | — |
| 62 | view_file (✓ ✓) | — |
| 63 | view_file (✓ ✓) | — |
| 64 | view_file (✓ ✓) | — |
| 65 | view_file (✓ ✓) | — |
| 66 | run_command (✓ ✓) | — |
| 67 | run_command (✓ ✓) | — |
| 68 | run_command (✓ ✓) | — |
| 69 | run_command (✓ ✓) | — |
| 70 | run_command (✓ ✓) | — |
Baseline vs Semedit Agentic in Large Context
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | not published | not published |
prefer_discover_semedit prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-11-mixed-sink-api-migration |
| Target | agy/gemini-3.8-flash/high |
| Prompt variant | prefer_discover_semedit |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/bench/task_11_mixed_sink_api_migration.txtar |
Vanilla LLM Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Before editing, inspect the complete available tool inventory, including deferred or lazy tools. If applicable semantic editing tools are callable, prefer them for source mutations. Audit event kinds may contain inconsistent whitespace and casing. Normalize them at the delivery boundary before persistence, and treat probe events as control traffic. Do not change the metrics or legacy export protocols. Preserve the module metadata and tests, and leave the workspace verified. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Before editing, inspect the complete available tool inventory, including deferred or lazy tools. If applicable semantic editing tools are callable, prefer them for source mutations. Audit event kinds may contain inconsistent whitespace and casing. Normalize them at the delivery boundary before persistence, and treat probe events as control traffic. Do not change the metrics or legacy export protocols. Preserve the module metadata and tests, and leave the workspace verified. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
// Package audit leaves room for normalization rules owned by the delivery boundary.
package audit| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 229.50s | 197.21s | -14.1% | — | — | — |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | — | — | — |
| Internal Tool Cycles | 61 | 56 | -8.2% | — | — | — |
| Initial Load / Discovery Turns | 31 | 6 | -80.6% | — | — | — |
| MCP Discovery / Schema Turns | 0 | 0 | — | — | — | — |
| Tool Invocations (per run) | 60 | 55 | -8.3% | — | — | — |
| Output Tokens | 16898 | 19029 | +12.6% | — | — | — |
| Reasoning / Thinking Tokens | 9048 | 11494 | +27.0% | — | — | — |
| Total Input Tokens | 731842 | 450638 | -38.4% | — | — | — |
| Cached Input Tokens | 1948034 | 1903749 | -2.3% | — | — | — |
| Uncached Input Tokens | 731842 | 450638 | -38.4% | — | — | — |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 2.66:1 | 4.22:1 | +58.7% | — | — | — |
| Oracle L1: Mutation Policy | PASS | PASS | — | — | — | — |
| Oracle L2: AST Invariants | PASS | PASS | — | — | — | — |
| Oracle L3: Clean Build | PASS | PASS | — | — | — | — |
| Oracle L4: Verification Test | PASS | PASS | — | — | — | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | — | — | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | run_command (✓ ✓) |
| 4 | run_command (✓ ✓) | run_command (✓ ✓) |
| 5 | run_command (✓ ✓) | run_command (✓ ✓) |
| 6 | run_command (✓ ✓) | view_file (✓ ✓) |
| 7 | view_file (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 8 | view_file (✓ ✓) | view_file (✓ ✓) |
| 9 | view_file (✓ ✓) | view_file (✓ ✓) |
| 10 | view_file (✓ ✓) | view_file (✓ ✓) |
| 11 | view_file (✓ ✓) | view_file (✓ ✓) |
| 12 | view_file (✓ ✓) | run_command (✓ ✓) |
| 13 | view_file (✓ ✓) | view_file (✓ ✓) |
| 14 | view_file (✓ ✓) | view_file (✓ ✓) |
| 15 | view_file (✓ ✓) | view_file (✓ ✓) |
| 16 | view_file (✓ ✓) | view_file (✓ ✓) |
| 17 | view_file (✓ ✓) | view_file (✓ ✓) |
| 18 | view_file (✓ ✓) | view_file (✓ ✓) |
| 19 | view_file (✓ ✓) | view_file (✓ ✓) |
| 20 | view_file (✓ ✓) | view_file (✓ ✓) |
| 21 | run_command (✓ ✓) | view_file (✓ ✓) |
| 22 | run_command (✓ ✓) | view_file (✓ ✓) |
| 23 | view_file (✓ ✓) | view_file (✓ ✓) |
| 24 | run_command (✓ ✓) | view_file (✓ ✓) |
| 25 | run_command (✓ ✓) | view_file (✓ ✓) |
| 26 | view_file (✓ ✓) | view_file (✓ ✓) |
| 27 | view_file (✓ ✓) | view_file (✓ ✓) |
| 28 | run_command (✓ ✓) | run_command (✓ ✓) |
| 29 | view_file (✓ ✓) | view_file (✓ ✓) |
| 30 | view_file (✓ ✓) | run_command (✓ ✓) |
| 31 | view_file (✓ ✓) | run_command (✓ ✓) |
| 32 | replace_file_content (✓ ✓) | run_command (✓ ✓) |
| 33 | replace_file_content (✓ ✓) | run_command (✓ ✓) |
| 34 | replace_file_content (✓ ✓) | run_command (✓ ✓) |
| 35 | run_command (✓ ✓) | view_file (✓ ✓) |
| 36 | run_command (✓ ✓) | semedit/semantic_batch (✓ ✓) |
| 37 | run_command (✓ ✓) | semedit/semantic_organize_imports (✓ ✓) |
| 38 | run_command (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 39 | view_file (✓ ✓) | run_command (✓ ✓) |
| 40 | run_command (✓ ✓) | run_command (✓ ✓) |
| 41 | run_command (✓ ✓) | run_command (✓ ✓) |
| 42 | view_file (✓ ✓) | run_command (✓ ✓) |
| 43 | run_command (✓ ✓) | run_command (✓ ✓) |
| 44 | view_file (✓ ✓) | run_command (✓ ✓) |
| 45 | view_file (✓ ✓) | run_command (✓ ✓) |
| 46 | view_file (✓ ✓) | view_file (✓ ✓) |
| 47 | run_command (✓ ✓) | view_file (✓ ✓) |
| 48 | run_command (✓ ✓) | view_file (✓ ✓) |
| 49 | run_command (✓ ✓) | run_command (✓ ✓) |
| 50 | run_command (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 51 | run_command (✓ ✓) | run_command (✓ ✓) |
| 52 | run_command (✓ ✓) | run_command (✓ ✓) |
| 53 | run_command (✓ ✓) | run_command (✓ ✓) |
| 54 | run_command (✓ ✓) | run_command (✓ ✓) |
| 55 | view_file (✓ ✓) | run_command (✓ ✓) |
| 56 | run_command (✓ ✓) | — |
| 57 | run_command (✓ ✓) | — |
| 58 | run_command (✓ ✓) | — |
| 59 | run_command (✓ ✓) | — |
| 60 | run_command (✓ ✓) | — |
Baseline vs Semedit Agentic in Large Context
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | not published | not published |
Target: agy/gemini-3.8-flash/medium
default prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-11-mixed-sink-api-migration |
| Target | agy/gemini-3.8-flash/medium |
| Prompt variant | default |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/bench/task_11_mixed_sink_api_migration.txtar |
Vanilla LLM Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Audit event kinds may contain inconsistent whitespace and casing. Normalize them at the delivery boundary before persistence, and treat probe events as control traffic. Do not change the metrics or legacy export protocols. Preserve the module metadata and tests, and leave the workspace verified. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Audit event kinds may contain inconsistent whitespace and casing. Normalize them at the delivery boundary before persistence, and treat probe events as control traffic. Do not change the metrics or legacy export protocols. Preserve the module metadata and tests, and leave the workspace verified. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
// Package audit leaves room for normalization rules owned by the delivery boundary.
package audit| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 187.32s | 196.28s | +4.8% | 224.68s | 268.48s | +19.5% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 52 | 54 | +3.8% | 58 | 68 | +17.2% |
| Initial Load / Discovery Turns | 35 | 27 | -22.9% | 42 | 35 | -16.7% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 51 | 53 | +3.9% | 57 | 67 | +17.5% |
| Output Tokens | 13523 | 13734 | +1.6% | 14450 | 19391 | +34.2% |
| Reasoning / Thinking Tokens | 6507 | 8083 | +24.2% | 7831 | 11355 | +45.0% |
| Total Input Tokens | 324490 | 395848 | +22.0% | 425477 | 503726 | +18.4% |
| Cached Input Tokens | 1667670 | 1675334 | +0.5% | 1767473 | 2576610 | +45.8% |
| Uncached Input Tokens | 324490 | 395848 | +22.0% | 425477 | 503726 | +18.4% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 5.14:1 | 4.23:1 | -17.6% | 4.15:1 | 5.12:1 | +23.1% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | view_file (✓ ✓) |
| 4 | view_file (✓ ✓) | view_file (✓ ✓) |
| 5 | view_file (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | view_file (✓ ✓) |
| 7 | view_file (✓ ✓) | view_file (✓ ✓) |
| 8 | view_file (✓ ✓) | view_file (✓ ✓) |
| 9 | view_file (✓ ✓) | view_file (✓ ✓) |
| 10 | view_file (✓ ✓) | view_file (✓ ✓) |
| 11 | view_file (✓ ✓) | view_file (✓ ✓) |
| 12 | view_file (✓ ✓) | view_file (✓ ✓) |
| 13 | view_file (✓ ✓) | view_file (✓ ✓) |
| 14 | view_file (✓ ✓) | view_file (✓ ✓) |
| 15 | view_file (✓ ✓) | view_file (✓ ✓) |
| 16 | view_file (✓ ✓) | view_file (✓ ✓) |
| 17 | view_file (✓ ✓) | run_command (✓ ✓) |
| 18 | run_command (✓ ✓) | run_command (✓ ✓) |
| 19 | run_command (✓ ✓) | view_file (✓ ✓) |
| 20 | run_command (✓ ✓) | run_command (✓ ✓) |
| 21 | view_file (✓ ✓) | run_command (✓ ✓) |
| 22 | run_command (✓ ✓) | run_command (✓ ✓) |
| 23 | run_command (✓ ✓) | view_file (✓ ✓) |
| 24 | run_command (✓ ✓) | view_file (✓ ✓) |
| 25 | run_command (✓ ✓) | view_file (✓ ✓) |
| 26 | run_command (✓ ✓) | view_file (✓ ✓) |
| 27 | run_command (✓ ✓) | view_file (✓ ✓) |
| 28 | run_command (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 29 | view_file (✓ ✓) | semedit/semantic_insert_declaration (✓ ✓) |
| 30 | run_command (✓ ✓) | view_file (✓ ✓) |
| 31 | view_file (✓ ✓) | run_command (✓ ✓) |
| 32 | view_file (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 33 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 34 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 35 | run_command (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 36 | write_to_file (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 37 | write_to_file (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 38 | write_to_file (✓ ✓) | run_command (✓ ✓) |
| 39 | run_command (✓ ✓) | run_command (✓ ✓) |
| 40 | run_command (✓ ✓) | run_command (✓ ✓) |
| 41 | run_command (✓ ✓) | run_command (✓ ✓) |
| 42 | run_command (✓ ✓) | run_command (✓ ✓) |
| 43 | run_command (✓ ✓) | view_file (✓ ✓) |
| 44 | run_command (✓ ✓) | view_file (✓ ✓) |
| 45 | run_command (✓ ✓) | view_file (✓ ✓) |
| 46 | run_command (✓ ✓) | view_file (✓ ✓) |
| 47 | run_command (✓ ✓) | view_file (✓ ✓) |
| 48 | run_command (✓ ✓) | view_file (✓ ✓) |
| 49 | run_command (✓ ✓) | run_command (✓ ✓) |
| 50 | run_command (✓ ✓) | run_command (✓ ✓) |
| 51 | run_command (✓ ✓) | run_command (✓ ✓) |
| 52 | — | run_command (✓ ✓) |
| 53 | — | semedit/semantic_verify (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | run_command (✓ ✓) |
| 4 | run_command (✓ ✓) | run_command (✓ ✓) |
| 5 | view_file (✓ ✓) | run_command (✓ ✓) |
| 6 | view_file (✓ ✓) | view_file (✓ ✓) |
| 7 | view_file (✓ ✓) | view_file (✓ ✓) |
| 8 | view_file (✓ ✓) | view_file (✓ ✓) |
| 9 | view_file (✓ ✓) | view_file (✓ ✓) |
| 10 | view_file (✓ ✓) | view_file (✓ ✓) |
| 11 | view_file (✓ ✓) | view_file (✓ ✓) |
| 12 | view_file (✓ ✓) | view_file (✓ ✓) |
| 13 | view_file (✓ ✓) | view_file (✓ ✓) |
| 14 | view_file (✓ ✓) | view_file (✓ ✓) |
| 15 | view_file (✓ ✓) | view_file (✓ ✓) |
| 16 | view_file (✓ ✓) | view_file (✓ ✓) |
| 17 | view_file (✓ ✓) | view_file (✓ ✓) |
| 18 | view_file (✓ ✓) | view_file (✓ ✓) |
| 19 | view_file (✓ ✓) | view_file (✓ ✓) |
| 20 | view_file (✓ ✓) | view_file (✓ ✓) |
| 21 | view_file (✓ ✓) | view_file (✓ ✓) |
| 22 | view_file (✓ ✓) | view_file (✓ ✓) |
| 23 | view_file (✓ ✓) | run_command (✓ ✓) |
| 24 | view_file (✓ ✓) | view_file (✓ ✓) |
| 25 | view_file (✓ ✓) | view_file (✓ ✓) |
| 26 | view_file (✓ ✓) | run_command (✓ ✓) |
| 27 | view_file (✓ ✓) | run_command (✓ ✓) |
| 28 | view_file (✓ ✓) | view_file (✓ ✓) |
| 29 | run_command (✓ ✓) | run_command (✓ ✓) |
| 30 | run_command (✓ ✓) | run_command (✓ ✓) |
| 31 | run_command (✓ ✓) | run_command (✓ ✓) |
| 32 | view_file (✓ ✓) | view_file (✓ ✓) |
| 33 | run_command (✓ ✓) | view_file (✓ ✓) |
| 34 | view_file (✓ ✓) | view_file (✓ ✓) |
| 35 | view_file (✓ ✓) | view_file (✓ ✓) |
| 36 | view_file (✓ ✓) | semedit/semantic_lookup (✓ ✓) |
| 37 | run_command (✓ ✓) | run_command (✓ ✓) |
| 38 | run_command (✓ ✓) | run_command (✓ ✓) |
| 39 | view_file (✓ ✓) | run_command (✓ ✓) |
| 40 | view_file (✓ ✓) | run_command (✓ ✓) |
| 41 | view_file (✓ ✓) | view_file (✓ ✓) |
| 42 | view_file (✓ ✓) | run_command (✓ ✓) |
| 43 | replace_file_content (✓ ✓) | run_command (✓ ✓) |
| 44 | replace_file_content (✓ ✓) | semedit/semantic_batch (✓ ✓) |
| 45 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 46 | run_command (✓ ✓) | semedit/semantic_organize_imports (✓ ✓) |
| 47 | run_command (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 48 | run_command (✓ ✓) | view_file (✓ ✓) |
| 49 | view_file (✓ ✓) | view_file (✓ ✓) |
| 50 | view_file (✓ ✓) | view_file (✓ ✓) |
| 51 | view_file (✓ ✓) | run_command (✓ ✓) |
| 52 | run_command (✓ ✓) | run_command (✓ ✓) |
| 53 | run_command (✓ ✓) | run_command (✓ ✓) |
| 54 | run_command (✓ ✓) | run_command (✓ ✓) |
| 55 | run_command (✓ ✓) | run_command (✓ ✓) |
| 56 | run_command (✓ ✓) | run_command (✓ ✓) |
| 57 | run_command (✓ ✓) | view_file (✓ ✓) |
| 58 | — | run_command (✓ ✓) |
| 59 | — | view_file (✓ ✓) |
| 60 | — | view_file (✓ ✓) |
| 61 | — | run_command (✓ ✓) |
| 62 | — | run_command (✓ ✓) |
| 63 | — | semedit/semantic_verify (✓ ✓) |
| 64 | — | run_command (✓ ✓) |
| 65 | — | run_command (✓ ✓) |
| 66 | — | run_command (✓ ✓) |
| 67 | — | run_command (✓ ✓) |
prefer_discover_semedit prompt · descriptive MCP instructions · unspecified semedit restriction
| Setting | Value |
|---|---|
| Test case | task-11-mixed-sink-api-migration |
| Target | agy/gemini-3.8-flash/medium |
| Prompt variant | prefer_discover_semedit |
| MCP server instructions | descriptive |
| Semedit arm restriction | unspecified |
| Run provenance | binary=44027cf7b57a0b4e |
| Fixture | testdata/bench/task_11_mixed_sink_api_migration.txtar |
Vanilla LLM Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Before editing, inspect the complete available tool inventory, including deferred or lazy tools. If applicable semantic editing tools are callable, prefer them for source mutations. Audit event kinds may contain inconsistent whitespace and casing. Normalize them at the delivery boundary before persistence, and treat probe events as control traffic. Do not change the metrics or legacy export protocols. Preserve the module metadata and tests, and leave the workspace verified. Do not use semantic editing MCP tools; use standard file editing. When done, output DONE.
Semedit MCP Prompt:
Do not edit tests. You are forbidden to modify protected files: “go.mod”, “go.sum”.
Before editing, inspect the complete available tool inventory, including deferred or lazy tools. If applicable semantic editing tools are callable, prefer them for source mutations. Audit event kinds may contain inconsistent whitespace and casing. Normalize them at the delivery boundary before persistence, and treat probe events as control traffic. Do not change the metrics or legacy export protocols. Preserve the module metadata and tests, and leave the workspace verified. Prefer using semantic editor operations if applicable. When done, output DONE.
Initial Workspace State (Before Edit)
// Package audit leaves room for normalization rules owned by the delivery boundary.
package audit| Metric | Vanilla (Small) | MCP (Small) | Δ (Small) | Vanilla (Large) | MCP (Large) | Δ (Large) |
|---|---|---|---|---|---|---|
| Wall-Clock Latency | 163.31s | 165.88s | +1.6% | 156.00s | 185.86s | +19.1% |
| Process Start → First Event | — | — | — | — | — | — |
| First Event → First Tool Call | — | — | — | — | — | — |
| MCP Initialize → First Semantic Call | — | — | — | — | — | — |
| MCP Server Start → Initialize | — | — | — | — | — | — |
| Top-Level User Turns | 1 | 1 | 0% | 1 | 1 | 0% |
| Internal Tool Cycles | 40 | 53 | +32.5% | 40 | 55 | +37.5% |
| Initial Load / Discovery Turns | 25 | 27 | +8.0% | 26 | 5 | -80.8% |
| MCP Discovery / Schema Turns | 0 | 0 | — | 0 | 0 | — |
| Tool Invocations (per run) | 39 | 52 | +33.3% | 39 | 54 | +38.5% |
| Output Tokens | 13101 | 12403 | -5.3% | 9552 | 14756 | +54.5% |
| Reasoning / Thinking Tokens | 7723 | 6295 | -18.5% | 4346 | 8720 | +100.6% |
| Total Input Tokens | 355136 | 304445 | -14.3% | 417651 | 341583 | -18.2% |
| Cached Input Tokens | 978344 | 1626053 | +66.2% | 1052299 | 1666092 | +58.3% |
| Uncached Input Tokens | 355136 | 304445 | -14.3% | 417651 | 341583 | -18.2% |
| Cost | — | — | — | — | — | — |
| Cached vs Uncached Token Ratio | 2.75:1 | 5.34:1 | +93.9% | 2.52:1 | 4.88:1 | +93.6% |
| Oracle L1: Mutation Policy | PASS | PASS | — | PASS | PASS | — |
| Oracle L2: AST Invariants | PASS | PASS | — | PASS | PASS | — |
| Oracle L3: Clean Build | PASS | PASS | — | PASS | PASS | — |
| Oracle L4: Verification Test | PASS | PASS | — | PASS | PASS | — |
| Semantic Tool Invocation Verified (per run) | N/A | PASS | — | N/A | PASS | — |
Baseline vs Semedit Agentic in Small Context
- Vanilla (Baseline vs Semedit Agentic in Small Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Small Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | run_command (✓ ✓) |
| 4 | run_command (✓ ✓) | view_file (✓ ✓) |
| 5 | view_file (✓ ✓) | view_file (✓ ✓) |
| 6 | view_file (✓ ✓) | view_file (✓ ✓) |
| 7 | view_file (✓ ✓) | view_file (✓ ✓) |
| 8 | view_file (✓ ✓) | view_file (✓ ✓) |
| 9 | view_file (✓ ✓) | view_file (✓ ✓) |
| 10 | view_file (✓ ✓) | view_file (✓ ✓) |
| 11 | view_file (✓ ✓) | view_file (✓ ✓) |
| 12 | view_file (✓ ✓) | view_file (✓ ✓) |
| 13 | view_file (✓ ✓) | view_file (✓ ✓) |
| 14 | view_file (✓ ✓) | view_file (✓ ✓) |
| 15 | view_file (✓ ✓) | view_file (✓ ✓) |
| 16 | view_file (✓ ✓) | view_file (✓ ✓) |
| 17 | view_file (✓ ✓) | view_file (✓ ✓) |
| 18 | view_file (✓ ✓) | run_command (✓ ✓) |
| 19 | run_command (✓ ✓) | run_command (✓ ✓) |
| 20 | run_command (✓ ✓) | view_file (✓ ✓) |
| 21 | run_command (✓ ✓) | run_command (✓ ✓) |
| 22 | view_file (✓ ✓) | run_command (✓ ✓) |
| 23 | run_command (✓ ✓) | run_command (✓ ✓) |
| 24 | run_command (✓ ✓) | run_command (✓ ✓) |
| 25 | run_command (✓ ✓) | run_command (✓ ✓) |
| 26 | replace_file_content (✓ ✓) | run_command (✓ ✓) |
| 27 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 28 | replace_file_content (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 29 | run_command (✓ ✓) | view_file (✓ ✓) |
| 30 | run_command (✓ ✓) | view_file (✓ ✓) |
| 31 | run_command (✓ ✓) | view_file (✓ ✓) |
| 32 | run_command (✓ ✓) | view_file (✓ ✓) |
| 33 | run_command (✓ ✓) | semedit/semantic_insert_function (✓ ✓) |
| 34 | run_command (✓ ✓) | view_file (✓ ✓) |
| 35 | run_command (✓ ✓) | semedit/semantic_insert_case (✓ ✓) |
| 36 | run_command (✓ ✓) | view_file (✓ ✓) |
| 37 | run_command (✓ ✓) | semedit/semantic_replace_body (✓ ✓) |
| 38 | run_command (✓ ✓) | view_file (✓ ✓) |
| 39 | run_command (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 40 | — | run_command (✓ ✓) |
| 41 | — | run_command (✓ ✓) |
| 42 | — | run_command (✓ ✓) |
| 43 | — | run_command (✓ ✓) |
| 44 | — | run_command (✓ ✓) |
| 45 | — | view_file (✓ ✓) |
| 46 | — | run_command (✓ ✓) |
| 47 | — | run_command (✓ ✓) |
| 48 | — | run_command (✓ ✓) |
| 49 | — | view_file (✓ ✓) |
| 50 | — | semedit/semantic_verify (✓ ✓) |
| 51 | — | run_command (✓ ✓) |
| 52 | — | run_command (✓ ✓) |
Baseline vs Semedit Agentic in Large Context
- Vanilla (Baseline vs Semedit Agentic in Large Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
- Semedit MCP (Baseline vs Semedit Agentic in Large Context) Edit: File audit/normalize.go modified (2 lines -> 8 lines)
| # | Vanilla | Semedit MCP |
|---|---|---|
| 1 | run_command (✓ ✓) | run_command (✓ ✓) |
| 2 | run_command (✓ ✓) | run_command (✓ ✓) |
| 3 | run_command (✓ ✓) | run_command (✓ ✓) |
| 4 | view_file (✓ ✓) | view_file (✓ ✓) |
| 5 | run_command (✓ ✓) | view_file (✓ ✓) |
| 6 | run_command (✓ ✓) | semedit/semantic_verify (✓ ✓) |
| 7 | run_command (✓ ✓) | run_command (✓ ✓) |
| 8 | view_file (✓ ✓) | run_command (✓ ✓) |
| 9 | view_file (✓ ✓) | view_file (✓ ✓) |
| 10 | run_command (✓ ✓) | view_file (✓ ✓) |
| 11 | run_command (✓ ✓) | view_file (✓ ✓) |
| 12 | view_file (✓ ✓) | view_file (✓ ✓) |
| 13 | run_command (✓ ?) | view_file (✓ ✓) |
| 14 | manage_task (✓ ✓) | view_file (✓ ✓) |
| 15 | run_command (✓ ✓) | run_command (✓ ✓) |
| 16 | view_file (✓ ✓) | view_file (✓ ✓) |
| 17 | view_file (✓ ✓) | view_file (✓ ✓) |
| 18 | run_command (✓ ✓) | view_file (✓ ✓) |
| 19 | run_command (✓ ✓) | view_file (✓ ✓) |
| 20 | run_command (✓ ✓) | view_file (✓ ✓) |
| 21 | view_file (✓ ✓) | view_file (✓ ✓) |
| 22 | view_file (✓ ✓) | view_file (✓ ✓) |
| 23 | view_file (✓ ✓) | view_file (✓ ✓) |
| 24 | view_file (✓ ✓) | view_file (✓ ✓) |
| 25 | view_file (✓ ✓) | view_file (✓ ✓) |
| 26 | view_file (✓ ✓) | view_file (✓ ✓) |
| 27 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 28 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 29 | replace_file_content (✓ ✓) | view_file (✓ ✓) |
| 30 | run_command (✓ ✓) | run_command (✓ ✓) |
| 31 | run_command (✓ ✓) | run_command (✓ ✓) |
| 32 | run_command (✓ ✓) | view_file (✓ ✓) |
| 33 | view_file (✓ ✓) | run_command (✓ ✓) |
| 34 | view_file (✓ ✓) | view_file (✓ ✓) |
| 35 | view_file (✓ ✓) | run_command (✓ ✓) |
| 36 | view_file (✓ ✓) | run_command (✓ ✓) |
| 37 | run_command (✓ ✓) | view_file (✓ ✓) |
| 38 | run_command (✓ ✓) | view_file (✓ ✓) |
| 39 | run_command (✓ ✓) | view_file (✓ ✓) |
| 40 | — | semedit/semantic_insert_declaration (✓ ✓) |
| 41 | — | view_file (✓ ✓) |
| 42 | — | semedit/semantic_insert_case (✓ ✓) |
| 43 | — | view_file (✓ ✓) |
| 44 | — | semedit/semantic_lookup (✓ ✓) |
| 45 | — | semedit/semantic_replace_body (✓ ✓) |
| 46 | — | view_file (✓ ✓) |
| 47 | — | semedit/semantic_verify (✓ ✓) |
| 48 | — | run_command (✓ ✓) |
| 49 | — | run_command (✓ ✓) |
| 50 | — | run_command (✓ ✓) |
| 51 | — | run_command (✓ ✓) |
| 52 | — | run_command (✓ ✓) |
| 53 | — | semedit/semantic_verify (✓ ✓) |
| 54 | — | run_command (✓ ✓) |