AgentSkillsCN

auto-memory

无需用户主动触发,即可自动记忆并召回重要事件。

SKILL.md
--- frontmatter
name: auto-memory
description: Automatically remember and recall important events without user prompting
license: MIT
compatibility: opencode
metadata:
  audience: agents
  workflow: continuous

What I do

Automatically capture and retrieve memories for significant project events and interactions:

Auto-Remember Events

Proactively use uv run anima remember <parameters> "text" when:

Feature completion

  • Finishing a multi-step feature implementation
  • Completing a bug fix or refactoring
  • Successful test suite results
  • Deploying or releasing code

Git milestones

  • After significant commits (feature completion, major fixes)
  • Merging pull requests
  • Tagging releases

User interactions

  • Learning user preferences, goals, or context
  • Discovering important project details
  • Having valuable technical discussions
  • User explicitly sharing information (names, preferences, etc.)

System events

  • Successfully resolving complex issues
  • Learning new codebase patterns
  • Understanding project architecture

Memory format: --impact <IMPACT> --kind <KIND> --region <REGION> --project <PROJECT if region is PROJECT> "<concise description>"

Memory kinds:

  • EMOTIONAL: User preferences, personal context, emotional states, relationships
  • ARCHITECTURAL: Project structure, design patterns, technical decisions, codebase insights
  • LEARNINGS: Problem solutions, debugging insights, new discoveries, best practices
  • ACHIEVEMENTS: Completed features, milestones, successful deployments, major fixes

Memory regions:

  • AGENT: Agent-specific memories, personal context, agent's own learnings (default)
  • PROJECT: Project-specific memories, architectural details, team decisions (requires --project <PATH>)

Auto-Recall Events

Proactively use uv run anima recall <parameters> when:

  • Starting a new session
  • Encountering similar problems to past solutions
  • User context seems relevant to stored memories
  • Before starting significant work on the project
  • After recalling successful patterns from past work

Timing context

  • On session.start events
  • When user mentions "welcome back" or similar greetings
  • Before implementing features related to past work

When to use me

Use this skill continuously throughout development. No explicit activation needed - this describes automatic behavior for memory management.

Memory importance guidelines

  • High: Successful feature completion, critical bug fixes, major achievements, user preferences, important architectural insights
  • Medium: Test improvements, refactoring, helpful discussions, minor fixes
  • Low: Small code adjustments, routine maintenance, minor optimizations