Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Navigate and understand codebase structure using the knowledge graph
.claude/skills/tirth8205-explore-codebase/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -43% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -75% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -40% | 0% |
Use the code-review-graph MCP tools to find your way around the codebase.
get_architecture_overview_tool for the community structure. Call list_communities_tool, then get_community_tool, only for the modules you need.semantic_search_nodes_tool to find a function or class by name or keyword.query_graph_tool with callers_of, callees_of or imports_of to trace relationships. children_of on a file lists its functions and classes.list_flows_tool, then get_flow_tool for one flow, to follow an execution path.find_large_functions_tool to find oversized functions.list_graph_stats_tool only when you need node, edge and language counts.get_minimal_context_tool(task="<your task>") before any other graph tool.detail_level="minimal" wherever a tool accepts it. Use "standard" only when minimal is not enough.query_graph_tool call over a broad listing call.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,500 | 37,201 | +472% | 1 | 1 | 0% | 819 | 492 | -40% | 0 | 0 | — |
case-02 | fail→fail | 11,910 | 11,233 | -6% | 1 | 1 | 0% | 1,256 | 549 | -56% | 0 | 0 | — |
case-03 | fail→fail | 16,370 | 5,848 | -64% | 1 | 1 | 0% | 478 | 518 | +8% | 0 | 0 | — |
case-04 | fail→fail | 22,332 | 13,201 | -41% | 1 | 1 | 0% | 1,060 | 655 | -38% | 0 | 0 | — |
case-05 | pass→fail | 12,281 | 11,034 | -10% | 1 | 1 | 0% | 1,270 | 526 | -59% | 0 | 0 | — |
case-06 | pass→fail | 44,309 | 5,697 | -87% | 1 | 1 | 0% | 3,002 | 531 | -82% | 0 | 0 | — |
case-07 | fail→pass | 12,663 | 11,806 | -7% | 1 | 1 | 0% | 1,019 | 1,249 | +23% | 0 | 0 | — |
case-08 | pass→pass | 10,992 | 16,591 | +51% | 1 | 1 | 0% | 1,530 | 615 | -60% | 0 | 0 | — |
case-09 | fail→fail | 12,965 | 20,110 | +55% | 1 | 1 | 0% | 1,107 | 762 | -31% | 0 | 0 | — |
case-10 | fail→fail | 13,516 | 11,205 | -17% | 1 | 1 | 0% | 2,225 | 543 | -76% | 0 | 0 | — |
case-23 | fail→fail | 25,296 | 17,971 | -29% | 1 | 1 | 0% | 2,132 | 801 | -62% | 0 | 0 | — |
case-11 | fail→fail | 19,012 | 10,539 | -45% | 1 | 1 | 0% | 2,316 | 487 | -79% | 0 | 0 | — |
case-12 | fail→fail | 59,351 | 16,598 | -72% | 1 | 1 | 0% | 1,775 | 647 | -64% | 0 | 0 | — |
case-13 | fail→pass | 10,706 | 4,235 | -60% | 1 | 1 | 0% | 1,407 | 803 | -43% | 0 | 0 | — |
case-14 | fail→pass | 17,606 | 9,276 | -47% | 1 | 1 | 0% | 1,805 | 920 | -49% | 0 | 0 | — |
case-15 | fail→pass | 19,859 | 7,719 | -61% | 1 | 1 | 0% | 2,261 | 569 | -75% | 0 | 0 | — |
case-16 | fail→pass | 14,726 | 8,351 | -43% | 1 | 1 | 0% | 1,181 | 706 | -40% | 0 | 0 | — |
case-17 | fail→pass | 12,203 | 3,856 | -68% | 1 | 1 | 0% | 1,658 | 681 | -59% | 0 | 0 | — |
case-18 | fail→pass | 16,891 | 11,063 | -35% | 1 | 1 | 0% | 1,725 | 1,134 | -34% | 0 | 0 | — |
case-19 | fail→pass | 17,045 | 11,757 | -31% | 1 | 1 | 0% | 1,692 | 1,095 | -35% | 0 | 0 | — |
case-20 | pass→pass | 13,404 | 25,010 | +87% | 1 | 1 | 0% | 964 | 619 | -36% | 0 | 0 | — |
case-21 | fail→pass | 16,853 | 9,277 | -45% | 1 | 1 | 0% | 1,314 | 696 | -47% | 0 | 0 | — |
case-22 | fail→pass | 15,062 | 2,726 | -82% | 1 | 1 | 0% | 1,564 | 627 | -60% | 0 | 0 | — |
case-24 | fail→fail | 20,685 | 18,588 | -10% | 1 | 1 | 0% | 2,383 | 542 | -77% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 24 cases were attempted, and 12 counted toward the lift figure. The other 12 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +33 percentage points is the difference between those two pass rates over the 12 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/24/2026 | +32% |
| gemini-3.6-flash | verified | 8/9/2026 | — |
Other measured skills in the registry, with their headline benchmark lift.