Skip to content
cccskills
FreePublished on GitHub

add-repo

Register a new sub-repo in the knowledge base.

The English summary is being prepared. Showing the original description.

See how to install

Files included (1)

  • SKILL.md3.3 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

add-repo

Adds a new sub-repo to the harness: updates local repos/repos.json (gitignored), clones, generates knowledge YAML.

If repos/repos.json is missing, copy repos/repos.json.example to repos/repos.json and edit entries there. Do not commit repos/repos.json.

When to Use

  • Bringing a new repository under Agent Forge management

Steps

1. Resolve location

Accept HTTPS, SSH URL, or local path. Extract repo name from URL (last segment without .git).

2. Check for duplicates

cat repos/repos.json | jq '.repos[] | select(.name == "<repo-name>")'

If already exists: report and stop.

3. Update repos.json

Add: {"name":"<name>","url":"<url>","defaultBranch":"main"}

For a dormant or reference-only repository, add "archived": true: it stays registered, and all-repos runs skip it unless it is named (see the syncing-repos skill, Archived repos).

4. Clone

bun run repo init --repo <repo-name> --human

5. Generate knowledge

/sync-knowledge <repo-name>

6. Give the quality gate the repository's checks

A Forge run that builds in this repository is gated from the harness (bun run quality-gate --correlation <checkout>/<pointer>). The gate runs the "checks" list on the repository's entry, in the run's checkout. Each check is a "name" and an "argv": the program, then its arguments. Ask the user which commands the repository's own checks are, and add them to the entry:

{
  "name": "my-service",
  "url": "https://github.com/org/my-service.git",
  "defaultBranch": "main",
  "checks": [
    { "name": "typecheck", "argv": ["bun", "run", "typecheck"] },
    { "name": "lint", "argv": ["bun", "run", "lint"] },
    { "name": "tests", "argv": ["bun", "test"], "timeoutSeconds": 900 }
  ]
}
  • Each check runs without a shell. There is no &&, no pipe, no $VAR and no glob: a check is one program with its arguments.
  • Windows: a check must name a real executable (an .exe). npm, pnpm, yarn and other .cmd shims can only be run by cmd.exe, so the gate refuses to run them and the check fails. Name node and the tool's script instead, for example ["node", "node_modules/typescript/bin/tsc", "--noEmit"], or the tool's own .exe.
  • A check's name is letters, digits, dot, dash or underscore, and must be unique. It may not be a check the gate runs itself: clean-tree, ac-verify, close-testing-attestation, test-evidence, eval-verdict or gate-log.
  • A check is stopped after two minutes unless it sets its own limit: "timeoutSeconds", a whole number of seconds from 1 to 3600 (one hour). Set it on a suite that takes longer. A check stopped at its limit fails the gate, and its entry says it was stopped for time ("timedOut": true, and an output that starts stopped after <n> seconds) rather than that it failed. When the gate runs as a hook, the host's own time limit for the hook still applies.
  • Ignore .tmp/ in the repository, in its .gitignore or in .git/info/exclude. The harness keeps the run's correlation and verdicts there. If the repository tracks files under .tmp/work/run-correlations or .tmp/work/evaluations, the gate refuses to run.
  • If the entry has no "checks" list, the gate refuses the run and names the entry to fill in.

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

Capture free-text work as a tracked Beads issue. Use when the user runs /add-bead or wants to quickly file a Beads issue.

The English summary is being prepared. Showing the original description.

jupes/agent-forge-harness2Updated Oct 22, 2026

Add focused Bun unit tests for mission-critical behavior and edge cases — not blanket coverage.

The English summary is being prepared. Showing the original description.

jupes/agent-forge-harness2Updated Oct 22, 2026

ask

Free

Answer questions about the codebase from knowledge files. Use when the user runs /ask or asks a domain/knowledge question about the repos.

The English summary is being prepared. Showing the original description.

jupes/agent-forge-harness2Updated Oct 22, 2026

Meta-skill for creating Agent Forge skills under .claude/skills/ with SKILL.md, optional references/ and scripts/, and Bun scaffolds. Use when the user wants to add or author a skill, scaffold a new skill folder, or align skill docs with harness conventions (JSON script output, Beads for tasks).

The English summary is being prepared. Showing the original description.

jupes/agent-forge-harness2Updated Oct 22, 2026

Choose Beads issue priority (P0–P4, numeric, or named) from urgency, impact, and risk when creating or triaging work.

The English summary is being prepared. Showing the original description.

jupes/agent-forge-harness2Updated Oct 22, 2026

Ultra-compressed communication mode that preserves technical accuracy while dropping filler. Use when the user explicitly says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman.

The English summary is being prepared. Showing the original description.

jupes/agent-forge-harness2Updated Oct 22, 2026

See all skills by jupes

Report a problem with this skill