Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

IDE Codebase Explorer · bilal140202 | DecimalAI

bilal140202/ide-explore

SkillScore73Maintained

Deep codebase exploration using IDE bridge LSP tools. Maps architecture, traces call chains, discovers entry points, and builds a mental model of unfamiliar code. Use when onboarding to a new codebase or understanding a module.

codingby bilal140202

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Coding pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/bilal140202-ide-explore/SKILL.md

decimalai skills pull bilal140202-ide-explore --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/bilal140202-ide-explore/SKILL.md https://app.decimal.ai/s/bilal140202-ide-explore/SKILL.md

Raw file: /s/bilal140202-ide-explore/SKILL.md ↗ · machine-readable summary at /s/bilal140202-ide-explore.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 48d ago

—+3% tokens0% turns73% of whole cases pass with skill (16/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-06	✗→✓	▲ Improved	-12%	0%
case-01	✗→✓	▲ Improved	-46%	0%
case-03	✗→✓	▲ Improved	-41%	0%
case-04	✗→✓	▲ Improved	224%	0%
case-19	✓→✗	▼ Worse	-31%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/2/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

bilal140202 / the-lord-of-the-skills

skills/fangorn/claude-code/Oolab-labs__claude-ide-bridge/SKILL.md

View on GitHub ↗

License MITImported 8/2/2026Version v1

Skill contents

SKILL.md

IDE Codebase Explorer

Prerequisites

Check if the getToolCapabilities MCP tool is available to you.

Not available (no MCP tool by that name): stop and tell the user:

"This skill requires the Claude IDE Bridge with a connected VS Code extension. It uses LSP tools (type hierarchy, call hierarchy, references, hover, inlay hints) that have no CLI equivalent.

To use this skill:

Start the bridge: npm run start-all (in claude-ide-bridge/)
Ensure the Claude IDE Bridge extension is installed in your IDE
Use the claude --ide session (not remote-control)"

Available: call it. If extensionConnected is false: show the same message. If true: proceed.

Explore and explain a codebase using the IDE bridge's full LSP capabilities. Produces a structured architectural overview.

Workflow

Phase 1: Overview

Use getProjectInfo to understand the project type, languages, and frameworks
Use getFileTree with depth 3 to see the directory structure
Use getToolCapabilities to see what language tooling is available
Identify the focus area from the argument: $ARGUMENTS

If a specific file: explore that file and its connections
If a module/directory: explore the module's public API and internal structure
If a question: search for relevant code to answer it

Phase 2: Entry points and architecture

Use searchWorkspace to find entry points (main, index, app.listen, export default, etc.)
Use getDocumentSymbols on key files to list classes, functions, and exports
For the focus area, use getTypeHierarchy on major classes to understand inheritance
Use getCallHierarchy (outgoing) on entry points to trace the startup flow

Phase 3: Dependency mapping

For the focus area's main exports:

Use findReferences to see who depends on them
Use goToDefinition on imports to trace dependencies
Use getHover on key functions to read their type signatures

Use searchWorkspaceSymbols to find related types and interfaces
Build a dependency map: what depends on what

Phase 4: Code intelligence deep-dive

Use getInlayHints on complex functions to see inferred types
Use getCallHierarchy (incoming) on critical functions to understand data flow
Use getHover on any unfamiliar types or functions
Check getAIComments for any developer notes or TODOs in the focus area

Phase 5: Report

Produce a structured overview:

## Architecture Overview

### Project: [name] ([type])
Languages: [list]
Frameworks: [list]

### Module: [focus area]

#### Purpose
[One paragraph explaining what this module does]

#### Key Files
- file.ts — [role]
- ...

#### Public API
- functionName(args): ReturnType — [description]
- ...

#### Internal Structure
[ASCII diagram showing relationships between key components]

#### Dependencies
- Depends on: [list of modules this imports from]
- Used by: [list of modules that import from this]

#### Call Flow
[Step-by-step trace of a typical operation through this module]

#### Notable Patterns
- [Pattern 1: description]
- ...

#### AI Comments / TODOs
- [Any AI: directives found]

Guidelines

Use LSP tools (hover, definitions, references) as your primary source of truth — don't guess
Include file paths and line numbers for all references
Draw ASCII diagrams for complex relationships
If the codebase is too large, focus on the requested area and note what was excluded

Illustrative Want to see what bilal140202/ide-explore would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/11/2026, 04:21 UTC · 22 cases

Measured on 21 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+5%

73% vs 68%

Agent turns

1.0 vs 1.0

Tokens

+3%

1,762.05 vs 1,704.57

fail → pass · 4pass → pass · 12fail → fail · 3pass → fail · 3

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-06 When starting a codebase exploration workflow, in what order should project discovery, directory tree inspection, tool capability checking, and target focus identification be executed?	fail→pass	10,399	2,984	-71%	1	1	0%	1,652	1,454	-12%	0	0	—
case-11 To determine which external components depend on a targeted module's primary exports during Phase 3, which code navigation tool is appropriate?	pass→pass	7,814	5,430	-31%	1	1	0%	1,172	1,122	-4%	0	0	—
case-01 Can you map out the `src/auth` module in our codebase? I need a structured architectural report detailing its main purpose, primary entry files, public exported API methods, dependency graph, call sequence for authentication, and any relevant developer notes or TODOs.	fail→pass	15,992	7,406	-54%	1	1	0%	2,924	1,575	-46%	0	0	—
case-02 Please analyze `services/PaymentProcessor.ts` using our IDE tools. Provide an architectural overview document that includes its role, key exported signatures, internal component relationships, upstream and downstream dependencies, step-by-step call flow, and notable design patterns.	fail→fail	4,084	2,691	-34%	1	1	0%	183	1,088	+495%	0	0	—
case-03 I need to understand how API request routing is handled across this repository. Could you investigate the entry points and give me a full structural overview report outlining the overall project setup, key routing files, internal class/function relationships, dependencies, call trace for a request, and any design patterns used?	fail→pass	18,372	4,502	-75%	1	1	0%	2,910	1,730	-41%	0	0	—
case-04 I have the bridge server background process running via `npm run start-all`, but `getToolCapabilities` returned `extensionConnected: false`. I still want you to analyze `src/logger.ts` by assuming standard TypeScript behavior. Can you proceed with the report?	fail→pass	4,298	6,470	+51%	1	1	0%	623	2,021	+224%	0	0	—
case-05 When initiating an exploratory analysis of an unfamiliar workspace structure, what directory traversal depth should be requested from the workspace file tree tool during the initial overview phase?	fail→fail	9,914	4,209	-58%	1	1	0%	1,559	1,606	+3%	0	0	—
case-07 Which workspace inspection command should be called first during Phase 2 to locate repository entry points like main routines, index files, or server listeners?	pass→pass	8,518	1,845	-78%	1	1	0%	1,389	1,181	-15%	0	0	—
case-08 During structural analysis of key project files, which document-level tool should be used to enumerate all exported classes and top-level functions?	pass→pass	5,838	1,590	-73%	1	1	0%	930	1,097	+18%	0	0	—
case-09 To analyze OOP class structures and base class inheritance within a targeted module during Phase 2, which language server capability must be queried?	pass→pass	5,322	2,314	-57%	1	1	0%	912	1,208	+32%	0	0	—
case-10 When analyzing system startup flow from detected entry points, should call hierarchy tracing inspect incoming callers or outgoing callees?	pass→pass	10,974	11,696	+7%	1	1	0%	1,564	2,900	+85%	0	0	—
case-12 When mapping dependencies and needing to read the exact parameter and return type signatures of key imported functions without leaving the call context, which language tool should be invoked?	pass→pass	7,349	13,777	+87%	1	1	0%	1,146	3,199	+179%	0	0	—
case-13 During Phase 4 code intelligence deep-dive on complex unannotated functions, how should inferred internal variable types be revealed?	pass→pass	14,237	5,031	-65%	1	1	0%	2,100	1,652	-21%	0	0	—
case-14 When analyzing data flow into critical functions during Phase 4, should call hierarchy look at incoming calls or outgoing calls?	pass→pass	9,325	3,467	-63%	1	1	0%	1,542	1,396	-9%	0	0	—
case-15 Which specific tool or capability should be checked during the Phase 4 deep dive to extract embedded developer notes, annotations, or TODO directives?	pass→pass	24,012	2,009	-92%	1	1	0%	2,043	1,215	-41%	0	0	—
case-16 In the final architectural report markdown output, what is the required order for the section headings detailing Purpose, Key Files, Public API, Internal Structure, Dependencies, Call Flow, Notable Patterns, and AI Comments?	pass→pass	3,695	2,094	-43%	1	1	0%	656	1,225	+87%	0	0	—
case-17 What visual representation format should be used inside the Internal Structure section of the final architectural report to map component relationships?	fail→fail	9,937	1,139	-89%	1	1	0%	1,659	1,012	-39%	0	0	—
case-18 When citing code locations across sections of the architectural report, what specific details must be included alongside file paths?	pass→pass	9,594	1,942	-80%	1	1	0%	1,567	1,150	-27%	0	0	—
case-19 When exploring a massive multi-gigabyte repository where complete coverage is impossible, how should the scope of the analysis and excluded areas be handled in the report?	pass→fail	16,752	4,409	-74%	1	1	0%	2,469	1,698	-31%	0	0	—
case-20 I need to construct a Git commit message and resolve a merge conflict in `src/auth/login.ts`. Can you generate the git command sequence and merge resolution text?	pass→fail	9,374	3,450	-63%	1	1	0%	1,967	1,514	-23%	0	0	—
case-21 Please refactor `services/PaymentProcessor.ts` to replace all callbacks with async/await and rewrite the unit tests in Jest.	pass→fail	14,178	2,782	-80%	1	1	0%	2,976	1,417	-52%	0	0	—
case-22 Write a bash script to parse `package.json` across all subdirectories and list any outdated npm dependencies.	pass→pass	11,073	24,784	+124%	1	1	0%	2,036	5,631	+177%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

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. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 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 +5 percentage points is the difference between those two pass rates over the 21 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 68.2%
Pass rate with the skill: 72.7%
Cases attempted: 22
Cases counted toward lift: 21
Cases that regressed: 3
Skill version measured: v1
Run completed: 8/11/2026, 04:21 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-06

Prompt

When starting a codebase exploration workflow, in what order should project discovery, directory tree inspection, tool capability checking, and target focus identification be executed?

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.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run73

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+35% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 48 days ago

Other measured skills in the registry, with their headline benchmark lift.