| name | project-agent-learning |
| description | Capture reviewable agent-learning findings for this repository after meaningful sessions. |
Project Agent Learning
Use this wrapper around agent-learning-consumer at task close-out, or when the user asks the agent to preserve a durable workflow lesson.
Fast Path
- Read this repository's learning root README and
infra/doc/agent-learning/config.json.
- Check existing memory, skills, docs, and constraints before writing a new finding.
- Capture only durable signals: user corrections, repeated failures, missed validation, repo rules, or hard-constraint candidates.
- Default first occurrences to
ADD_LEARNING_NOTE; use IGNORE for praise, summaries, and one-off task details.
- Require
pattern_key and validation_case.given / when / then for every non-IGNORE learning.
- Run
vendor/bin/agent-learning validate --root infra/doc/agent-learning.
- Hand off durable guidance edits to
project-agent-guidance-maintenance; do not self-approve promotions.
Commands
vendor/bin/agent-learning validate --root infra/doc/agent-learning
vendor/bin/agent-learning prepare --root infra/doc/agent-learning --task PROJECT-123
vendor/bin/agent-learning proposal-validate --root infra/doc/agent-learning --proposal proposal.2026-06-15.001.json