Skip to content

README.md

Latest commit

 

History

History
72 lines (45 loc) · 2.52 KB

README.md

File metadata and controls

72 lines (45 loc) · 2.52 KB

Continual Learning

Automatically and incrementally keeps AGENTS.md up to date from transcript changes.

The plugin combines:

  • A stop hook that decides when to trigger learning.
  • A continual-learning skill that orchestrates the learning flow.
  • An agents-memory-updater subagent that mines new or changed transcripts and updates AGENTS.md.

It is designed to avoid noisy rewrites by:

  • Reading existing AGENTS.md first and updating matching bullets in place.
  • Processing only new or changed transcript files.
  • Writing plain bullet points only (no evidence/confidence metadata).

Installation

/add-plugin continual-learning

How it works

On eligible stop events, the hook prefers launching the agent CLI in print mode when agent is available on PATH. If the CLI is unavailable or cannot be started, the hook falls back to emitting a followup_message that asks the current session to run the continual-learning skill.

The skill is marked disable-model-invocation: true, so it will not be auto-selected during normal model invocation. When it does run, it delegates the full memory update flow to agents-memory-updater.

The hook keeps local runtime state in:

  • .cursor/hooks/state/continual-learning.json (cadence state)
  • .cursor/hooks/state/continual-learning-agent.log (best-effort agent CLI output when launched from the hook)

The updater uses an incremental transcript index at:

  • .cursor/hooks/state/continual-learning-index.json

Trigger cadence

Default cadence:

  • minimum 10 completed turns
  • minimum 120 minutes since the last run
  • transcript mtime must advance since the previous run

Trial mode defaults (enabled in this plugin hook config):

  • minimum 3 completed turns
  • minimum 15 minutes
  • automatically expires after 24 hours, then falls back to default cadence

Optional env overrides

  • CONTINUAL_LEARNING_MIN_TURNS (or legacy CONTINUOUS_LEARNING_MIN_TURNS)
  • CONTINUAL_LEARNING_MIN_MINUTES (or legacy CONTINUOUS_LEARNING_MIN_MINUTES)
  • CONTINUAL_LEARNING_TRIAL_MODE (or legacy CONTINUOUS_LEARNING_TRIAL_MODE)
  • CONTINUAL_LEARNING_TRIAL_MIN_TURNS (or legacy CONTINUOUS_LEARNING_TRIAL_MIN_TURNS)
  • CONTINUAL_LEARNING_TRIAL_MIN_MINUTES (or legacy CONTINUOUS_LEARNING_TRIAL_MIN_MINUTES)
  • CONTINUAL_LEARNING_TRIAL_DURATION_MINUTES (or legacy CONTINUOUS_LEARNING_TRIAL_DURATION_MINUTES)

Output format in AGENTS.md

The memory updater writes only:

  • ## Learned User Preferences
  • ## Learned Workspace Facts

Each item is a plain bullet point.

License

MIT