forget

Retract outdated or contradictory memory records after user confirmation.

3|1|Updated Dec 9, 2025
One-click install
npx skills add https://github.com/isizono/calm --skill forget-isizono
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: forget
Source: https://github.com/isizono/calm/tree/main/skills/forget
Command: npx skills add https://github.com/isizono/calm --skill forget-isizono

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Long-running AI-assisted projects accumulate decisions, habits, tag notes, and rules that eventually contradict current direction. This Skill detects stale or conflicting records and guides a safe, user-confirmed retraction process instead of letting obsolete context silently mislead future sessions. ## Core Features & Use Cases - Autonomous trigger detection: Activates when a newly recorded decision conflicts with past records, when the user signals a policy reversal, or when stale habits and rules are noticed during work. - Storage-aware retraction: Applies the correct removal method per record type, including logical retraction for decisions, logs, and materials, deactivation for habits, full-text rewrite for tag notes, and physical deletion for auto-memory, CLAUDE.md, rules files, and skills. - Dependency checking for skills: Greps other skill definitions and hook configurations before deleting a skill to surface references for user judgment. - Use Case: After recording a new decision that reverses an earlier design choice, the Skill flags the old decision as a retraction candidate, presents it with rationale, and retracts it only after explicit approval. ## Quick Start Ask the assistant to review past records that contradict the current direction and retract the outdated ones after confirmation.

Frequently Asked Questions about forget

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I retract an outdated decision recorded in a previous session?▼

Present the conflicting record as a retraction candidate with its rationale, then execute the retract tool only after the user approves that specific item. Retraction is a logical deletion, so the record history remains auditable.

What triggers automatic detection of stale memory records?▼

Detection triggers include recording a new decision that contradicts past decisions or habits, user phrasing signaling a policy reversal, noticing drift between old rules and current practice, and finding cleanup targets when closing a discussion.

Can habits and tag notes be retracted the same way as decisions?▼

No, habits are deactivated via update_habit with active set to false because the retract tool does not accept them. Tag notes have no retract concept, so the tag is rewritten with the outdated portion removed.

Does deleting a skill check for references from other skills first?▼

Yes, before physically deleting a skill, the skill name is grepped across other SKILL.md files and hooks configuration. Any references found are shown to the user, who makes the final decision on whether to proceed.

When should retraction not be performed?▼

Retraction is skipped for open discussion points without conclusions, when the user explicitly declined it, or when judgment is uncertain. False-positive retractions risk losing important records, so doubtful cases default to keeping the record.