yaggo-brain-community

Shared collective-memory pack for yaggo-brain.

This repository is a repo-as-hub: a single git-tracked JSON file (community-lessons.json) that holds generalized, reusable lessons learned by AI coding agents. Any yaggo-brain instance can pull it over the network, and anyone can contribute back via a pull request. No server, no accounts, no cost — the network effect runs on git.

What is a "lesson"?

A lesson is a small, generalized learning distilled from a failure or recurring pattern — never project-specific code. Each entry:

{
  "symptom": "Short description of the observed problem",
  "cause": "Why it happens (root cause)",
  "fix": "The actionable, testable remedy",
  "appliesTo": { "tags": ["category", "topic"], "language": "optional" },
  "submissionId": "optional-id"
}

Rules for a good lesson:

• Generalized: no file paths, secrets, tokens, emails, or company-specific detail.
• Actionable: the fix is concrete and testable.
• Scoped: use appliesTo.tags (and optional language/framework) so instances can re-rank by stack.

Consume it (pull community memory)

Point your yaggo-brain instance at this pack with HUB_URL, then update:

# .env
HUB_URL=https://raw.githubusercontent.com/YaggoSEO/yaggo-brain-community/main/community-lessons.json

# In your yaggo-brain checkout:
npx yaggo-brain update      # pulls HUB_URL into packages/hub/seed/community-lessons.json
pnpm db:migrate             # loads the pack into your database (idempotent)

yaggo-brain update uses HUB_URL when set (this raw JSON), then falls back to a local API, then to the bundled seed pack. Loading on db:migrate means every clone bootstraps the shared memory automatically — local-first and private.

Contribute a lesson

Option A — Pull request (recommended):

1. Fork this repo.
2. Add your entry to community-lessons.json (keep it generalized; see rules above).
3. Open a PR. Maintainers review for quality, privacy, and duplication.

Option B — From your yaggo-brain instance:

1. npx yaggo-brain contrib --lesson-id=<uuid> (requires HUB_OPT_IN=true). This generalizes the lesson, runs eval gates, and writes it to your local pack.
2. Copy the accepted entry into a PR here.

Moderation

Contributions are reviewed before merge. We reject entries that contain PII/secrets, are too project-specific, are not actionable, or duplicate an existing lesson.

License

Released under CC0-1.0 (public domain dedication) — reuse freely. See LICENSE.