Skill Claude CodeCodex
Develop/iterate on the MidMem middleware core (this repo's packages/core — the shared "LLM Wiki" knowledge layer). Use when adding or changing MidMem capability — retrieval lanes, tiers/lifecycle, work-memory events, concept routing, claims, grounding, the MCP/CLI/hook surfaces, or its tests. Encodes the…
Skill Claude CodeCodex
Ingest LLM Wiki knowledge AND review/audit its quality and the two stacks' understanding — a check-and-balance over the shared MidMem store. Use for "ingest and review", "review/audit the knowledge base", "knowledge quality check", "what does OpenClaw vs Hermes understand about X", "is our stored knowledge…
Skill Claude CodeCodex
Ingest a document, URL, or a recurring research digest into MidMem with grounding verification. Use for "ingest this", "ingest the weekly report/paper/URL", or committing a research digest to the knowledge base. Verifies the grounding report instead of blind-ingesting, stages URLs into an allowed source root, and…
Skill Claude CodeCodex
Orchestrate MidMem (LLM Wiki) knowledge-curation pipelines with the frontier-plans / Hermes-builds / frontier-QA loop — bulk ingestion, re-grounding, dedup/supersede cleanup, tier-promotion review, vault verification. Use for "curate the knowledge base", "bulk ingest these docs", "rebuild/clean the wiki", "orchestrate…
Skill Claude CodeCodex
Capture architecture/coding patterns, reusable scaffolding, anti-patterns and recipes into MidMem via capture packs — at the moment a build proves them. Use after a build/QA gate passes ("record this pattern", "capture the scaffold", "remember this anti-pattern") or when a repeated approach keeps working across…
Skill Claude CodeCodex
Record a change/decision/lesson durably into MidMem the RIGHT way — a distilled lesson into the wisdom tier, a clean commit, and (optionally) a changelog entry. Use after any capability/config change or when asked to "record this", "log it", "remember this lesson", "checkpoint". Encodes the harness-guaranteed…
Skill Claude CodeCodex
Digest a recurring research report (e.g. a weekly arXiv review of memory-systems papers) end-to-end: verified ingest into MidMem, a GROUNDED gap-analysis of every recommendation against the actual MidMem source code (not memory of it), a published digest for the operator, and roadmap deltas recorded in the repo. Wraps…