Menu
Walk through stale or conflicting memory entries and routing-table drift. Use when the user runs /sustain:audit, asks "is my memory store healthy?", or before a major release of personal rules.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Basé sur la classification professionnelle SOC
| name | audit-memory |
| description | Walk through stale or conflicting memory entries and routing-table drift. Use when the user runs /sustain:audit, asks "is my memory store healthy?", or before a major release of personal rules. |
This skill processes the report produced by /sustain:audit and helps the user act on it. Never auto-deletes anything — the user reviews every change.
/sustain:audit runs two scans and prints a JSON-ish report:
Memory audit (lib/audit/memory.js → audit()):
stale[] — drawers not modified in > 90 daysduplicates[] — pairs of drawers with similar slugs in the same wingurls[] — URLs extracted from drawers (not fetched — verify yourself if suspect)Routing audit (lib/audit/routing.js → audit(spec)):
listedButMissing[] — spec.skillRouting references a skill that isn't installedinstalledButUnrouted[] — installed skill that no routing entry mentionsFor each item in stale:
Read tool, the path field).utime) to reset the clock, or just acknowledge.Don't ask one question per drawer for sessions with many stale entries — batch into a list and let the user mark them in one message.
For each pair in duplicates:
Extract dead-URL candidates from urls. Suspect signals:
localhost, 127.0.0.1, internal hosts that may have moved.Ask the user before fetching — WebFetch has a cost and the user may already know which ones are dead.
For listedButMissing:
claude-mem:smart-explore is an MCP search tool, not a filesystem skill). Check the notes field of the audit report.spec.json.For installedButUnrouted:
using-superpowers itself).spec.json and ask the user whether the scenario fits their workflow.After processing, suggest the user re-run /sustain:audit to confirm the report is empty (or has only known-acceptable noise). If the user has mempalace newly installed, mention lib/audit/migrate.js → buildPlan() for one-shot fs → mempalace import.
Search persistent memory across mempalace MCP, claude-mem MCP, and the structured filesystem fallback. Use when the user asks "how did we solve X before?", "what did we decide about Y?", or anything that implies recalling prior-conversation context.
Decide whether and where to persist a learning. Use at phase boundaries when something seems worth long-term memory — a rule, a non-obvious decision, an external resource, or a project-state change. Picks the right wing (feedback / project / reference / user) and routes through the user's active memory backend.
Interpret session token statistics and recommend the next move — /compact, delegate to a subagent, switch to a smaller model, or stop and persist learnings. Use after `/sustain:status`, when the session feels long, or when context-window pressure is showing up in degraded responses.
Run the 6-question phase-end checklist when a logical phase completes (after a feature, before a context switch, or when the Stop hook prompts you). Use to surface skipped Iron Rules, missed memory writes, and overdue /compact opportunities.
Establish Iron Rules (3+1) and R1–R5 token-saving rules at session start or whenever the rules feel forgotten. Use when user asks about token efficiency, when rules need re-grounding, or proactively at the start of complex multi-turn work.