How do I install the Repository Archaeology skill?

Run npx skillmds@latest add jbaehne/repository-archaeology in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Repository Archaeology safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Repository Archaeology?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Repository Archaeology free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Repository Archaeology?

jbaehne (@jbaehne) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Repository Archaeology by jbaehne — Agent Skill

Name: Repository Archaeology
Rating: 4.2 (5 reviews)
Author: jbaehne

Repository Archaeology

Discover repository structure, local instructions, canonical commands, and the Git change boundary before planning or editing. Use when entering a repository, when conventions are unknown, or when asked to map the repo or inspect working-tree state. Do not use to implement changes, write the implementation plan, or rule on merge readiness.

jbaehne Updated Sep 22, 2026

File contents

Repository archaeology

Establish what this repository actually is before planning or editing. Report observed facts. Mark inferences and unknowns. Do not invent paths, tools, branches, or commands.

Repository-local instructions override this skill. If those instructions conflict, or the canonical place to change cannot be determined, stop and state the smallest decision needed.

Inspect before concluding

Read the instructions that exist. Do not assume a file is missing until you have looked. Typical sources, when present:

AGENTS.md, CLAUDE.md, Cursor rules, and other agent instruction files
contribution guides, README, skill or coding standards
package manifests, task runners, CI config, and existing validation scripts
templates and tests that show the real layout

Then inspect the tree, Git status, and recent history far enough to name:

the skill or code layout this repo actually uses
build, test, lint, and validation commands that are documented or scripted
the diff boundary for the requested work
unrelated working-tree changes, which you must leave untouched
architectural patterns that constrain the change
uncertainties that would change the plan

Prefer commands the repository already documents. Do not install tools or run formatters as a side effect of discovery.

Do not

treat this survey as a plan or as permission to edit
modify files, including to "tidy" unrelated dirty paths
claim a command succeeded unless you ran it and saw the result
assume GitHub or GitLab, or a language, from habit

Output

Lead with the map, not a transcript.

## Repository map
- Purpose and canonical locations (paths you confirmed)

## Applicable instructions
- Path and the rule that overrides generic defaults
- Conflicts, if any

## Commands
- Documented or scripted: `command` — source
- Not established: what you still do not know

## Git
- Revision, branch, and status
- Unrelated dirty or untracked paths to preserve

## Patterns that constrain the change
- Observed pattern — where you saw it

## Uncertainties
- Item that would change planning or editing

Omit empty sections. Separate what you opened or ran from what you inferred.

When handing off, fill only known fields from ../_sdlc/evidence.yaml. Set state.discovered only after this survey. Leave state.planned and state.authorized false.

Suite boundaries: ../_sdlc/LIFECYCLE.md.

---name: repository-archaeologydescription: Discover repository structure, local instructions, canonical commands, and the Git change boundary before planning or editing. Use when entering a repository, when conventions are unknown, or when asked to map the repo or inspect working-tree state. Do not use to implement changes, write the implementation plan, or rule on merge readiness.---# Repository archaeologyEstablish what this repository actually is before planning or editing. Report observed facts. Mark inferences and unknowns. Do not invent paths, tools, branches, or commands.Repository-local instructions override this skill. If those instructions conflict, or the canonical place to change cannot be determined, stop and state the smallest decision needed.## Inspect before concludingRead the instructions that exist. Do not assume a file is missing until you have looked. Typical sources, when present:- `AGENTS.md`, `CLAUDE.md`, Cursor rules, and other agent instruction files- contribution guides, README, skill or coding standards- package manifests, task runners, CI config, and existing validation scripts- templates and tests that show the real layoutThen inspect the tree, Git status, and recent history far enough to name:- the skill or code layout this repo actually uses- build, test, lint, and validation commands that are documented or scripted- the diff boundary for the requested work- unrelated working-tree changes, which you must leave untouched- architectural patterns that constrain the change- uncertainties that would change the planPrefer commands the repository already documents. Do not install tools or run formatters as a side effect of discovery.## Do not- treat this survey as a plan or as permission to edit- modify files, including to "tidy" unrelated dirty paths- claim a command succeeded unless you ran it and saw the result- assume GitHub or GitLab, or a language, from habit## OutputLead with the map, not a transcript.```markdown## Repository map- Purpose and canonical locations (paths you confirmed)## Applicable instructions- Path and the rule that overrides generic defaults- Conflicts, if any## Commands- Documented or scripted: `command` — source- Not established: what you still do not know## Git- Revision, branch, and status- Unrelated dirty or untracked paths to preserve## Patterns that constrain the change- Observed pattern — where you saw it## Uncertainties- Item that would change planning or editing```Omit empty sections. Separate what you opened or ran from what you inferred.When handing off, fill only known fields from [`../_sdlc/evidence.yaml`](../_sdlc/evidence.yaml). Set `state.discovered` only after this survey. Leave `state.planned` and `state.authorized` false.Suite boundaries: [`../_sdlc/LIFECYCLE.md`](../_sdlc/LIFECYCLE.md).

Frequently asked questions