Claude Skill

rabbithole-doc-exploration

Use local Rabbithole for human-in-the-loop exploration of AGENTS.md, SKILL.md, SET plans, repomaps, and long repo docs. Use when the user wants to inspect, question, branch, or review dense agent-facing documents interactively before changing durable docs or workflow contracts.

LLM Mart · 0 points · 2 views 0 listing impressions 0 install-command copies
Download markoblogo-abvx-agent-skills-skills_rabbithole-doc-exploration-37a7146.zip · 2 KB

Install

skills CLI npx skills add https://github.com/markoblogo/abvx-agent-skills/tree/main/skills/rabbithole-doc-exploration
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install markoblogo-abvx-agent-skills@llmmart
Git git clone https://github.com/markoblogo/abvx-agent-skills.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole markoblogo/abvx-agent-skills collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Rabbithole Doc Exploration

Use a local Rabbithole MCP canvas to inspect dense documents with the human before turning findings into durable repo changes.

Use When

  • reviewing AGENTS.md, RUNBOOK.md, docs/ai/repomap.compact.md, or generated repo contracts;
  • reviewing SKILL.md files, skill cards, or workflow packs;
  • reviewing SET planner exports such as orchestrator-bundle.json or rabbithole.seed.md;
  • the user wants to branch into questions from selected text instead of reading a long document linearly.

Inputs

Good starting documents:

  • docs/ai/rabbithole.seed.md
  • AGENTS.md
  • RUNBOOK.md
  • docs/ai/repomap.compact.md
  • .set-plan/orchestrator-bundle.json
  • .set-plan/rabbithole.seed.md
  • skills/<name>/SKILL.md

Workflow

  1. Check availability
    • If Rabbithole MCP tools are available, use open_rabbithole with a title and file path or content.
    • If they are not available, continue as a normal document review and tell the user Rabbithole is optional.
  2. Open the smallest useful seed
    • Prefer docs/ai/rabbithole.seed.md or .set-plan/rabbithole.seed.md when present.
    • Otherwise open the single document the user named.
  3. Answer branches
    • Treat each selected passage as the anchor.
    • Answer with concrete file paths, risks, and proposed doc/code follow-ups.
    • Keep answers branch-sized; do not turn every branch into a full audit.
  4. Promote findings carefully
    • Only write durable repo docs when the finding is stable and broadly useful.
    • Keep Rabbithole notes out of CI contracts unless the repo explicitly stores them.
  5. Close with a compact synthesis
    • List decisions, doc changes made, open questions, and any branch findings that should become tasks.

Guardrails

  • Do not make Rabbithole a required runtime dependency for generated docs, skills, or SET.
  • Do not treat Rabbithole canvas state as source of truth.
  • Do not paste secrets, private profiles, or sensitive local files into Rabbithole unless the user explicitly approves.
  • Prefer seed docs over huge raw files.

Fallback

If Rabbithole is unavailable, read the target document directly and produce the same synthesis without the interactive canvas.

Files (abvx-agent-skills)
  • agents
    • openai.yaml 252 B
      Loading…
  • SKILL.md 2.6 KB
    ---
    name: rabbithole-doc-exploration
    description: Use local Rabbithole for human-in-the-loop exploration of AGENTS.md, SKILL.md, SET plans, repomaps, and long repo docs. Use when the user wants to inspect, question, branch, or review dense agent-facing documents interactively before changing durable docs or workflow contracts.
    license: MIT
    metadata:
      abvx_status: experimental
      abvx_origin: original
    ---
    
    # Rabbithole Doc Exploration
    
    Use a local Rabbithole MCP canvas to inspect dense documents with the human before turning findings into durable repo changes.
    
    ## Use When
    
    - reviewing `AGENTS.md`, `RUNBOOK.md`, `docs/ai/repomap.compact.md`, or generated repo contracts;
    - reviewing `SKILL.md` files, skill cards, or workflow packs;
    - reviewing SET planner exports such as `orchestrator-bundle.json` or `rabbithole.seed.md`;
    - the user wants to branch into questions from selected text instead of reading a long document linearly.
    
    ## Inputs
    
    Good starting documents:
    
    - `docs/ai/rabbithole.seed.md`
    - `AGENTS.md`
    - `RUNBOOK.md`
    - `docs/ai/repomap.compact.md`
    - `.set-plan/orchestrator-bundle.json`
    - `.set-plan/rabbithole.seed.md`
    - `skills/<name>/SKILL.md`
    
    ## Workflow
    
    1. **Check availability**
       - If Rabbithole MCP tools are available, use `open_rabbithole` with a title and file path or content.
       - If they are not available, continue as a normal document review and tell the user Rabbithole is optional.
    2. **Open the smallest useful seed**
       - Prefer `docs/ai/rabbithole.seed.md` or `.set-plan/rabbithole.seed.md` when present.
       - Otherwise open the single document the user named.
    3. **Answer branches**
       - Treat each selected passage as the anchor.
       - Answer with concrete file paths, risks, and proposed doc/code follow-ups.
       - Keep answers branch-sized; do not turn every branch into a full audit.
    4. **Promote findings carefully**
       - Only write durable repo docs when the finding is stable and broadly useful.
       - Keep Rabbithole notes out of CI contracts unless the repo explicitly stores them.
    5. **Close with a compact synthesis**
       - List decisions, doc changes made, open questions, and any branch findings that should become tasks.
    
    ## Guardrails
    
    - Do not make Rabbithole a required runtime dependency for generated docs, skills, or SET.
    - Do not treat Rabbithole canvas state as source of truth.
    - Do not paste secrets, private profiles, or sensitive local files into Rabbithole unless the user explicitly approves.
    - Prefer seed docs over huge raw files.
    
    ## Fallback
    
    If Rabbithole is unavailable, read the target document directly and produce the same synthesis without the interactive canvas.
    
  • SKILL_CARD.md 1.4 KB
    Loading…

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.