feat: add /learn command to extract session learnings to scoped AGENTS.md files

[R44VC0RP]

Summary

• Adds /learn built-in slash command
• Analyzes session context and extracts non-obvious learnings
• Creates/updates AGENTS.md files at the appropriate directory level (not just root)
• Leverages the new scoped AGENTS.md feature from feat: dynamically resolve AGENTS.md files from subdirectories as agent explores them #10678

Usage

/learn

The command will review the session for discoveries, determine the right scope for each learning, and update the appropriate AGENTS.md file(s).

[github-actions]

Thanks for your contribution!

This PR doesn't have a linked issue. All PRs must reference an existing issue.

Please:

1. Open an issue describing the bug/feature (if one doesn't exist)
2. Add Fixes #<number> or Closes #<number> to this PR description

See CONTRIBUTING.md for details.

[github-actions]

The following comment was made by an LLM, it may be inaccurate:

No duplicate PRs found

[rekram1-node]

Maybe we should dogfood first by putting in

.opencode/commands/learn.md

Also thinking, we may wanna make the wording more about:

• read over the corrections I brought up to you, how could we tweak the relevant AGENTS.md instructions to prevent this confusion in the future? Use the most minimal possible prompt to ..... etc etc

[rekram1-node]

Assuming I understood the intent of the command?

[rekram1-node]

I'd actually use this but it doesnt hurt to dogfood b4 we add it to everyones configs