
Spec Miner
Reads existing code and writes up what it actually does as structured requirements
- Repository
- Jeffallan/claude-skills
- Stars
- 12,176
- Licence
- MIT
- Files
- 5 files
What it does
This skill teaches an AI agent to act as a reverse-engineering specialist for code you inherited or that has no documentation. Instead of guessing, the agent searches and reads the project's files, traces how data and requests move through it, and writes a specification describing the behaviour it observed, using a structured 'shall' style called EARS. It is meant to produce a written report saved to a specs folder, with uncertainties flagged separately from confirmed facts.
Great for
- Developers taking over a legacy or undocumented codebase
- Teams who need API or architecture documentation generated from source code
- Engineers planning changes to an existing feature who need to know its current behaviour
Skip it if
- Projects where you already have good documentation or a formal requirements source, since the skill only reflects what the code does, not what it should do
- Codebases you cannot share with an AI tool, because the agent reads the source code directly
- Anyone expecting a verified or complete spec without human review, since the output is inferred from code and can miss intent or runtime behaviour
Try asking
- “Reverse engineer this legacy Django project and document its API endpoints”
- “I inherited this old codebase with no docs, figure out how the order processing works”
- “Create architecture documentation from the implementation in this repo”
- “Identify undocumented business logic in the billing module”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 5 original files, the MIT licence and an attribution note.
spec-miner.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip spec-miner.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip spec-miner.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Point the agent at a specific feature or module first, since whole-system scans on large repos can be long and shallow
- Review the uncertainties section carefully and answer open questions with the original team or domain experts before relying on the spec
- Ask the agent to cite file and line locations for each requirement so you can verify them quickly
What's inside
- references/analysis-checklist.md
- references/analysis-process.md
- references/ears-format.md
- references/specification-template.md
- SKILL.md
Redistributed unmodified from Jeffallan/claude-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.