agentsclimarketplace
Sign in

Project graph context

Skill DmitrySelenya/safe-project-graph/skills/project-graph-context

Use when reviewing code, assessing a local diff or blast radius, finding related files or tests, exploring symbols, or preparing non-trivial edits and refactors with project graph context.From its SKILL.md

Install
npx -y skills add DmitrySelenya/safe-project-graph --skill project-graph-context

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 19 days oldThe repository was created 19 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.8 KB, 587 tokens by cl100k_base, as published. Nobody here has run it

Project Graph Context

Use the bundled wrapper for advisory structural evidence. Always verify graph results by reading the real source files and tests.

Resolve the wrapper

Locate this SKILL.md through the skill path exposed by the agent client. Change to the directory containing it, then invoke the bundled script by its relative path:

cd "<skill-root>"
python3 scripts/project_graph.py --help

Replace <skill-root> with the actual directory containing this file. Never hardcode a home directory or assume the repository contains the skill.

Loading this file is the activation. Do not activate another same-named skill or switch skill roots; duplicate installations may contain different wrappers.

Workflow

  1. Record git -C <repo> status --porcelain so the starting worktree state can be compared later.
  2. Run python3 scripts/project_graph.py doctor --repo <repo>.
  3. Run python3 scripts/project_graph.py setup only when doctor reports that the managed runtime is missing.
  4. Run build --repo <repo> when the graph database is missing. Otherwise run update --repo <repo> --base HEAD.
  5. Query only what the task needs:
python3 scripts/project_graph.py impact --repo <repo> --base HEAD
python3 scripts/project_graph.py find <symbol-or-path> --repo <repo>
python3 scripts/project_graph.py related <symbol-or-path> --repo <repo>

find and related accept --limit (default 20, capped at 100). Raise it only when a broad query returns a truncated list.

  1. Read the actual files and tests surfaced by the graph before editing or reporting conclusions.
  2. Re-run git -C <repo> status --porcelain and confirm it matches the starting state.

Every command fails with a single ERROR: line and a non-zero exit code. Report that line as-is; do not retry blindly. If it says the graph database is missing, has no table, or lacks columns, run build --repo <repo> once and retry.

Safety boundaries

  • Use only this wrapper. Do not run upstream install, hooks, serve, watch, daemon, refactor/apply, embeddings, or wiki commands.
  • Keep indexes in the wrapper-managed external cache. Never create .code-review-graph* inside the repository.
  • If doctor reports repository-local graph artifacts, stop and report their exact paths. Do not remove them automatically.
  • Do not edit agent instructions, permissions, hooks, MCP configuration, or repository settings as part of graph setup.
  • Treat risk, test-gap, flow, and relationship output as advisory evidence, not proof.

Use tentative language: “The graph suggests…”, followed by the source and test evidence used to verify it.

What ships with it: 2 files

24.4 KB alongside SKILL.md, 1 of them executable

agents/

scripts/

Keep looking

Skills are one crate of 326,645. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.