omcr-setup

name

omcr-setup

description

Install OMCR infrastructure in a project — scaffold empty CLAUDE.md marker blocks (`## Project context` / `## Research stack` / `## Language preference`), create `.claude/agent-memory/` for the 6 core agents with canonical-template MEMORY.md, initialize empty `references.bib` + `references.csv`, and install a curated permission allowlist in `.claude/settings.json`. Does NOT ask about the user's research project (no working title / hypothesis / venue / preset interview). For that, run `/start-research` after this completes. Safe to re-run — never overwrites existing user content.

Setup

Install-style flow that lays down the infrastructure OMCR needs to operate in a project. This is OMCR's analog of OMC's /omc-setup — no research-interview questions, no preset selection, no manuscript scaffolding. Just makes the plugin work.

After /omcr-setup completes, the user should run /start-research for interview-style first-project initialization.

When this skill is invoked, immediately execute the workflow below. Do not only restate or summarize these instructions back to the user.

Safe to re-run. Diffs against current state; only adds what is missing.

What this installs

1. CLAUDE.md infrastructure — ## Project context, ## Research stack, ## Language preference blocks with placeholder fields. (Filled in later by /start-research.)
2. Agent memory directories — .claude/agent-memory/<agent>/MEMORY.md for the 6 core agents, seeded from the canonical template. Existing files are never touched.
3. Bibliography files — empty paper/references.bib (with header comment) and references.csv (with header row) at default paths. Existing files are never touched.
4. Permission allowlist — curated .claude/settings.json so common research-flow tools (read-only git, file search, edits, LaTeX build, citation lookups, figure crop) don't prompt every time. Dangerous operations (git write, file deletion, wildcard bash) are intentionally excluded so they always prompt.

What this explicitly does NOT do

• Ask about your project (working title, hypothesis, target venue, datasets, etc.)
• Pick or apply a domain preset (neuro-fmri, etc.)
• Scaffold the LaTeX manuscript directory
• Clone Overleaf or commit anything

All of those belong to /start-research.

Phase execution

Execute phases sequentially. For each phase, read the linked file and follow its instructions exactly.

1. Phase 1 — State check. Read phases/01-state-check.md.
2. Phase 2 — CLAUDE.md scaffold. Read phases/02-claude-md-scaffold.md.
3. Phase 3 — Agent memory. Read phases/03-agent-memory.md.
4. Phase 4 — Bibliography. Read phases/04-bibliography.md.
5. Phase 5 — Permissions. Read phases/05-permissions.md.
6. Phase 6 — Report. Read phases/06-report.md.

Re-running policy

• CLAUDE.md marker blocks are inserted only if missing. Existing blocks are inspected but not overwritten.
• Existing MEMORY.md files are never overwritten (phase 3).
• Existing references.bib / references.csv are never overwritten (phase 4).
• Permission allowlist (phase 5) prompts the user before replacing or merging. Existing broad wildcards trigger a narrowing offer with backup to .claude/settings.json.backup.YYYY-MM-DD.

Final recommendation

After phase 6's report, recommend the user run /start-research to fill in their project context and scaffold the manuscript.