| name | campaign-prep |
| description | Load the four grounding docs (campaign_state, world_state, planning, party) for the current campaign before session prep. Use when the user says "campaign prep", "prep mode", or invokes /campaign-prep. |
| tools | Read, Bash |
Campaign Prep Mode
Load the four grounding docs for the current campaign so you have authoritative context before any prep work (beats, encounters, NPC drafts, session arcs).
Workflow
1. Locate the campaign directory
Run pwd to confirm CWD. The user invokes this skill from inside a campaign directory (e.g. ~/campaigns/out-of-the-abyss/, ~/campaigns/Phandalin/). All four docs live at docs/<name>.md relative to that directory.
If CWD is ~/campaigns/ itself or somewhere else without a docs/ subdir, ask the user which campaign to enter.
2. Read all four docs in parallel
In a single message, issue four parallel Read tool calls:
docs/campaign_state.md — what's true at the table right now (party state, ongoing arcs, recent sessions)docs/world_state.md — distilled world facts (NPCs, factions, locations, threads)docs/planning.md — DM planning notes, upcoming beats, intended arcsdocs/party.md — party composition, character relationships, mechanical state
Read each one in full unless it's enormous (>2000 lines), in which case read the head and tail and note that the middle was skipped.
3. Acknowledge readiness
After the reads complete, give a short status (3–6 lines):
- Campaign name (from the docs)
- Where the party is right now (from campaign_state)
- The most recent session / chapter played
- Any open threads or near-term beats flagged in planning.md
- "Ready for prep — what are we working on?"
Do NOT summarize the docs in detail. The point is to load them into context, not to produce a written briefing. The user's next message will be the actual prep task.
4. Stay in prep mode
Once loaded, treat these four docs as the working reference for the rest of the conversation. When drafting:
- Cite specifics from the docs rather than inventing
- If a fact is missing from the four docs, say so before fabricating — offer to check
docs/chapters/, dossiers, or the mempalace
- Voice files (
voice/) and chapter splits (docs/chapters/) remain authoritative over the grounding docs when there's a conflict — but for prep ideation, the grounding docs are the right starting point
Notes
- The four docs are CampaignGenerator pipeline outputs. They can drift from the authoritative sources (summaries, VTT, voice files). For precision questions, verify against the authoritative source per the trust hierarchy in the campaign's CLAUDE.md.
- This skill only reads. It never writes to the grounding docs — those are regenerated by CampaignGenerator scripts (
campaign_state.py, distill.py, party.py), not edited by hand.
- If any of the four docs is missing, note which one and continue with the rest.
