engrAIm

Set up a brand-new EngrAIm workspace for a folder/project. Use on first run (the SessionStart nudge), when the user runs /engraim:onboard, or when adopting EngrAIm into an existing project. Scans the project deterministically, seeds the memory frame (ENV/USER/SOUL), offers history backfill and optional semantic recall, and explains the workflow — without turning it into an interrogation.

Record and manage how Claude should BEHAVE in this workspace — corrections, conventions, do/don't rules, tone. Use WHENEVER the user corrects you, states a preference, expresses frustration with an approach, or a `verdict:` annotation comes back negative. Also use to reaffirm, promote, or retire existing behavior overrides. This is the workspace's self-improving frame; it's distinct from facts about the world (use engraim-memory for those).

Distill session transcripts into durable memory. Use when processing staged sessions (after the Stop hook), when the user runs /engraim:curate or /engraim:backfill, or whenever you should turn "what just happened" into lasting knowledge. Turns raw Claude Code session JSONL into wiki pages + temporal facts, with provenance and no secrets.

How to use the workspace's persistent memory. ALWAYS use this whenever you need to recall what was learned about THIS project/environment before, or whenever you learn something durable worth keeping. Use it for: "what do we know about X", recalling prior decisions, conventions, environment quirks, fixes, inventory; and for recording new durable facts. Drives the `engraim-memory` MCP tools (recall, remember, whats_true, wiki_get, wiki_upsert). Prefer recalling from memory before searching the web or asking the user to re-explain context.

The research-as-intake reflex. Use this WHENEVER you hit a knowledge gap or an error you can't resolve from memory — instead of guessing or giving up, search the web, prefer canonical sources, solve the problem, and intake the verified finding into the workspace knowledge base. Trigger on: unfamiliar errors, "how do I do X", version-specific behavior, tool/config questions about this environment, or anything where memory came up empty.

Run a periodic, GUARDED retrospective over the workspace memory: consolidate learnings, resolve verdicts, prune stale overrides, surface promotions. Use when the user runs /engraim:retro, when status shows many pending sessions or expiring overrides, or when a body of work has accumulated and the memory needs tightening. Enforces an echo-chamber guardrail and an action test so the retrospective improves the memory rather than just reinforcing it.

Author, review, and promote workspace-local skills so this project can grow its own capabilities. Use when a multi-step procedure recurs across sessions, when the user says "make this a skill" / "automate this", or when running /engraim:draft-skill, /engraim:promote-skill, or /engraim:skills. Drives the draft -> gate -> activate pipeline (draft_skill, review_skill, promote_skill, list_skills, retire_skill). Composes with Anthropic's skill-creator when it's available rather than duplicating it.