agentleFS
eduardtomasekeduardtomasek/agent-code-intel/.agents/skills/agent-code-intel-routing/SKILL.md

agent-code-intel-routing

Skill0 starsChanged 2 months agoMITView on GitHub

Automatically choose between GrepAI, GitNexus, and ripgrep for codebase discovery, architecture exploration, debugging, reviews, refactoring, and implementation tasks.

The file

Show its rules and text

We split this skill into the rules it gives an agent

22 rules. About 0 are common, 20 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈20

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill as the source of truth for choosing a code-intelligence tool.

    line 10
  2. Very specific:

    Before reading a source file, use the repo-scoped code-context skill.

    line 17
  3. Distinctive:

    Use semantic search for intent and behavior:

    line 29
  4. Distinctive:

    Use the repository graph for structure and relationships:

    line 37
  5. Distinctive:

    Run impact analysis before changing shared or unfamiliar symbols and before large refactors.

    line 44
  6. Very specific:

    Use rg for exact text and final verification:

    line 49
  7. Distinctive:

    Start with ripgrep when the target string or file is already known.

    line 55
  8. Distinctive:

    Use GrepAI to discover the relevant domain when its location is unknown.

    line 61
  9. Distinctive:

    Use GitNexus to inspect relationships, dependencies, and execution flows.

    line 62
  10. Distinctive:

    Read the relevant implementation before drawing conclusions.

    line 63
  11. Distinctive:

    Use GrepAI to locate the behavior behind the symptom.

    line 67
  12. Distinctive:

    Use GitNexus to trace the dependency or call chain.

    line 68
  13. Distinctive:

    Read the relevant code, identify the root cause, and verify it with exact searches or tests.

    line 69
  14. Distinctive:

    Use GrepAI to locate the implementation area when it is not already known.

    line 74
  15. Distinctive:

    Use GitNexus to understand surrounding architecture and impact.

    line 75
  16. Distinctive:

    Read affected files, implement the smallest change, then verify with tests and ripgrep.

    line 76
  17. Distinctive:

    Use GitNexus to determine blast radius and inspect references.

    line 81
  18. Distinctive:

    Read every affected implementation.

    line 82
  19. Distinctive:

    Verify remaining references with ripgrep and run relevant tests.

    line 84
  20. Distinctive:

    Search and graph results identify where to read.

    line 88
  21. Distinctive:

    When both indexes are unavailable, use ripgrep and direct file reads.

    line 96
  22. Distinctive:

    Treat an unavailable tool as a fallback condition.

    line 98

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md
<!-- agent-code-intel:managed -->

Code Intelligence Routing

Use this skill as the source of truth for choosing a code-intelligence tool. Choose the smallest tool that answers the current question; combine tools only when the task crosses their boundaries. Preserve repository safety rules such as required GitNexus impact analysis, graph-aware renames, and change detection.

Code-context rules

Before reading a source file, use the repo-scoped code-context skill.

These rules apply regardless of the rest of this routing:

  • Exhaustive reference lists for cross-module calls belong to rg/ast-grep, not the knowledge graphs.
  • Derive the exact definition range; never guess it or truncate it.

Tool roles

GrepAI

Use semantic search for intent and behavior:

  • locating business logic without knowing its identifiers;
  • discovering where a feature or responsibility lives;
  • exploring domain concepts and likely change locations.

GitNexus

Use the repository graph for structure and relationships:

  • dependency and module relationships;
  • architecture and execution flows;
  • callers, callees, impact analysis, and refactor blast radius;
  • graph-aware symbol renames and affected-flow verification.

Run impact analysis before changing shared or unfamiliar symbols and before large refactors.

ripgrep

Use rg for exact text and final verification:

  • identifiers, config keys, environment variables, migrations, and logs;
  • scripts and documentation;
  • confirming that references or stale text remain after a change.

Start with ripgrep when the target string or file is already known.

Workflows

Architecture and behavior

  1. Use GrepAI to discover the relevant domain when its location is unknown.
  2. Use GitNexus to inspect relationships, dependencies, and execution flows.
  3. Read the relevant implementation before drawing conclusions.

Debugging

  1. Use GrepAI to locate the behavior behind the symptom.
  2. Use GitNexus to trace the dependency or call chain.
  3. Read the relevant code, identify the root cause, and verify it with exact searches or tests.

Feature implementation

  1. Use GrepAI to locate the implementation area when it is not already known.
  2. Use GitNexus to understand surrounding architecture and impact.
  3. Read affected files, implement the smallest change, then verify with tests and ripgrep.

Refactoring

  1. Use GitNexus to determine blast radius and inspect references.
  2. Read every affected implementation; use graph-aware rename support for symbol renames.
  3. Verify remaining references with ripgrep and run relevant tests.

Evidence gate

Search and graph results identify where to read; they are not implementation evidence by themselves. Read the relevant source before editing or concluding.

Fallbacks

  • Without GrepAI, use GitNexus for discovery and ripgrep for exact search.
  • Without GitNexus, use GrepAI for discovery, direct source reads for relationships, and ripgrep for verification.
  • When both indexes are unavailable, use ripgrep and direct file reads.

Treat an unavailable tool as a fallback condition. Retry it only when the task is to repair that tool.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.