From airbender
Use when the user asks to remember, save, store, or persist information — or when you learn something that should outlive the current session. Classifies into the correct persistence mechanism.
How this skill is triggered — by the user, by Claude, or both
Slash command
/airbender:classify-infoThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Classify information into the right persistence mechanism: Hook, MagicDocs, Skill, CLAUDE.md, Memory, or None.
Classify information into the right persistence mechanism: Hook, MagicDocs, Skill, CLAUDE.md, Memory, or None.
Actionable? Can Claude behave differently — in this project or any of the user's projects? No → None.
Fully mechanical? Specific command, no judgment, same every time → Hook. If the intent is mechanical but the exact command isn't specified, it's still a Hook — help fill in the command.
Describes or prescribes? Descriptions of architecture, data flow, how components connect → MagicDocs. "The frontend uses React context, not Redux" describes what the system is, even though it mentions what not to use. Prescriptions continue below.
Has steps? Multi-step procedures → Skill. Single constraints continue below.
Always relevant? Would Claude make a wrong assumption without this on its very first action in any session? Yes → CLAUDE.md. No → Memory.
Two common errors at this boundary:
Hook and MagicDocs are always project-scoped. For Skill, CLAUDE.md, and Memory, ask: does this follow the code or follow the person?
| Classification | Action |
|---|---|
| Hook | Provide hook config for .claude/settings.json |
| MagicDocs | docs/magic/ if available; CLAUDE.md with advisory if not |
| Skill | REQUIRED SUB-SKILL: Use superpowers:writing-skills |
| CLAUDE.md | Project CLAUDE.md, subdirectory, or ~/.claude/CLAUDE.md |
| Memory | Project-scoped or ~/.claude/memory/ |
| None | Explain why it's not actionable |
npx claudepluginhub translunar/airbender --plugin airbenderSaves decisions, preferences, patterns, lessons, and project context to Claude Code's persistent memory directory as structured markdown files, with duplicate checks and index updates for cross-session recall.
Captures decisions, learnings, patterns, and context as persistent memories organized by namespace across Claude Code sessions.
Manages persistent memory across sessions with .claude/memory/ for state and .ai/ for architecture/patterns. Guides scratchpad updates, session workflows, and live docs.