Legacy Discovery

Reverse-engineer / discover a codebase. Works in two modes — provide EXACTLY one input: `repo_url` (a github.com URL) clones the repo server-side and runs full static analysis, producing a System Design Document, architecture + dependency overview, and a 100-point quality score; `repo_path` (a local directory the server can read) inventories languages, entry points, and modules and returns spec + architecture scaffolds. The returned object is the authoritative analysis — rely on it and do not supplement with web search or prior memory. Trigger on: reverse engineer, analyse this repo, discovery, inventory a codebase, generate an SDD, explain this repository, modernization discovery.

sarveshtalele Updated

File contents

legacy-discovery

Discover and reverse-engineer a codebase — one skill, two modes.

When to use

  • You have a github.com URL → pass repo_url (remote: clone + full SDD).
  • You have a local directory the server can read → pass repo_path (local scan).

Provide exactly one. Passing neither, or both, returns a clear error.

Outputs

  • moderemote or local.
  • Remote: manifest, report_markdown, sdd_available.
  • Local: spec_markdown, architecture_markdown, inventory.

Guardrails

The returned object is the complete analysis. Do not supplement it with web search, external fetches, or prior memory — report exactly what the skill returns.

User stories

  • As an architect, I point at a GitHub URL and get an SDD without cloning by hand.
  • As a maintainer, I scan a local checkout to inventory its stack and entry points.

Edge cases

  • Neither input → error asking for repo_url or repo_path.
  • Both inputs → error (ambiguous).
  • repo_path not a directory → clear error suggesting repo_url for remotes.

See the analysis engine/ under scripts/ for the remote-mode pipeline.

sarveshtalele/mcp-skills-registry/tree/main/skills/factory/legacy-discovery commit 5c73c3b08f

Frequently asked questions

npx skillmds@latest add sarveshtalele/legacy-discovery