managing-memories
Managing Memories
Extract persistent memories at natural breakpoints, not mid-chain.
When to extract
Extract when ALL conditions are met:
- Volume threshold: substantial conversation has occurred (roughly 10+ back-and-forth exchanges or significant work completed)
- Natural pause: task completed, waiting for user input, or session ending
- Novelty: something worth remembering that isn't already captured
- No pending work: the last assistant turn has no in-progress tool calls or unfinished chains
Do NOT extract:
- Mid-chain (during active multi-step implementation or debugging)
- After trivial exchanges (quick questions, one-liners, status checks)
- When the information is already in project instruction files or existing memories
- When the conversation is primarily about information retrieval (reading files, answering questions) with no decisions or corrections
Memory types
More from beltonk/claude-code-agent-skills
coding-practices
Coding-specific practices for AI agents — scope discipline, read-before-write, simplest approach first, incremental development, verification, comment standards, security awareness, tool preferences, and shell discipline. Use when the agent is writing, editing, or reviewing code. Works alongside agentic-standards (which covers general behavior for all interactions).
4scaffolding-projects
Provides a structured approach to starting new features or projects. Guides the agent through understanding requirements, exploring existing code, planning, incremental implementation, and verification. Use when asked to build something new — a feature, module, service, or project — to avoid jumping into code without context.
4handing-off-sessions
Captures structured session state for resuming work in a new session or handing off to another agent. Use at the end of a session, before context limits, or when the user asks to save progress. Not needed for trivial sessions (quick questions, one-line answers).
4compacting-context
Provides a structured 9-section summarization template for compressing long conversations while preserving critical details. Use when a session approaches context limits and history must be compressed without losing user intent, file changes, errors, or next steps.
4agentic-standards
Foundational behavioral standards for any AI agent — safety/reversibility framework, output quality, memory conventions, and prompt injection defense. Applies to all agent interactions including chat, analysis, writing, debugging, and coding. Use when setting up an agent, onboarding to a new project, or when behavioral baseline guidance is needed. For coding-specific rules, also load coding-practices.
4receiving-code-review
Enforces rigorous handling of code review feedback. Classifies comments by severity, verifies claims independently, prevents blind implementation of incorrect suggestions, and handles contradictory reviews. Use when receiving review comments on a pull request or code change.
1