agentleFS
MesteriisMesteriis/Engineering-Bible-AI/skills/architecture-map/SKILL.md

architecture-map

Skill0 starsChanged 4 months agoMITView on GitHub

Maps verified domains, dependencies, entrypoints, flows and risks for broad changes or review.

We split this skill into the rules it gives an agent

14 rules. 0 are common, 11 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 11

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read repo instructions and top-level dependency/build files.

    line 12
  2. Distinctive:

    Keep the output short enough to guide implementation or review.

    line 16
  3. Distinctive:

    When Graphify is exposed in the current environment and the task is about architecture, dependencies, call graphs, knowledge graphs, or …

    line 20
  4. Distinctive:

    Before building anything, check for existing state:

    line 25
  5. Very specific:

    If graphify-out/graph.json exists, query it first for architecture, dependency, and call-flow questions.

    line 31
  6. Distinctive:

    If no graph exists, initialize Graphify only when the map will benefit from a knowledge graph: broad architecture, dependency/call graph, …

    line 37
  7. Very specific:

    Do not create graphify-out/ for tiny localized tasks.

    line 47
  8. Distinctive:

    Treat Graphify output as an index, not as final proof.

    line 51
  9. Very specific:

    If Graphify is unavailable or not useful for the repository shape, fall back to rg, git ls-files, targeted file reads, and top-level …

    line 55
  10. Distinctive:

    When Graphify state is created or reused, report whether it was existing, fresh, stale/uncertain, or newly generated.

    line 58
  11. Distinctive:

    Use Archify as an optional renderer when a shareable architecture or process diagram helps the task and the current runtime is verified.

    line 63
  12. Distinctive:

    No renderer installation or diagram generation is implied by this skill alone.

    line 77
  13. Distinctive:

    test and validation commands found

    line 85
  14. Distinctive:

    Do not invent architecture.

    line 91

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “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

Architecture Map

Build a compact, verified architecture map from current files.

Workflow

  1. Read repo instructions and top-level dependency/build files.
  2. Identify entrypoints, layers, data flow, external integrations, and tests.
  3. Map dependency direction and boundary violations.
  4. Record uncertainty separately from verified facts.
  5. Keep the output short enough to guide implementation or review.

Graphify Route

When Graphify is exposed in the current environment and the task is about architecture, dependencies, call graphs, knowledge graphs, or repeated onboarding to a large codebase, use Graphify as the preferred Tier 2 knowledge-graph source.

Before building anything, check for existing state:

test -f graphify-out/graph.json && echo graph-ready

If graphify-out/graph.json exists, query it first for architecture, dependency, and call-flow questions. Check whether relevant files changed after the graph was built, or whether available Graphify metadata points at a different root/scope. If freshness is uncertain, use the graph as a hypothesis source and verify against files.

If no graph exists, initialize Graphify only when the map will benefit from a knowledge graph: broad architecture, dependency/call graph, repeated onboarding, large subsystem exploration, or multi-agent handoff. Prefer the narrowest useful scope:

graphify extract . --code-only
graphify extract <subdir> --code-only

Do not create graphify-out/ for tiny localized tasks. For read-only review, security review, validation-critical work, or an explicit no-mutation request, ask before creating Graphify state.

Treat Graphify output as an index, not as final proof. Verify important architecture claims against repository files, command output, or explicit user context before reporting them.

If Graphify is unavailable or not useful for the repository shape, fall back to rg, git ls-files, targeted file reads, and top-level dependency/build files.

When Graphify state is created or reused, report whether it was existing, fresh, stale/uncertain, or newly generated.

Optional Diagram Rendering

Use Archify as an optional renderer when a shareable architecture or process diagram helps the task and the current runtime is verified. Build its structured input from source-backed facts after mapping; keep source paths and revision evidence alongside the diagram. Graphify discovers candidate relationships; a renderer presents the supplied relationships.

Validate the structured input and rendered artifact, then visually inspect labels, edges and layout. A valid diagram does not prove a relationship exists in code, runtime impact, security, or merge safety. Keep uncertain links marked as hypotheses. Do not add rendering to ordinary localized edits.

For offline/private work, disable the renderer's update check through the documented runtime option. Keep generated HTML, snapshots and machine paths outside tracked files unless the user requests a documentation artifact. No renderer installation or diagram generation is implied by this skill alone.

Output

  • repo shape
  • main layers or bounded domains
  • important entrypoints
  • persistence/integration/config surfaces
  • test and validation commands found
  • risks and unknowns
  • suggested next skill if applicable

Rule

Do not invent architecture. Every claim must come from files, commands, or explicit user-provided context.

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.