outline-runbook-curator
Skill2 starsChanged 6 months agoMITView on GitHub
Curate, consolidate, and restyle the StumpCloud Runbooks wiki in Outline against the repo as source of truth. Use this skill whenever the user wants to review/clean up the wiki, consolidate or reorganize runbooks, fix stale or duplicated docs, reconcile Outline against the repo (ADRs, inventory, docs/), restructure the doc tree into hierarchy-native nodes, or apply the domain color + topical icon scheme to wiki nodes. Trigger on phrases like "audit the runbooks", "clean up the wiki", "consolidate the Outline docs", "the wiki is out of date", "reorganize/restyle the Runbooks collection", or "color and icon the docs". This is the governance/curation counterpart to the `stumpcloud-omg` skill (which files individual incidents); use that for postmortems, this for whole-collection curation. Always propose a consolidation plan and get approval before any destructive merge/move/delete.
In this repository7 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- cut-releaseSkill
- gemini-mockupSkill
- pr-sweepSkill
- refresh-miatrix-tokenSkill
- retirement-planSkill
- self-reportSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 19
Keep the StumpCloud Runbooks wiki accurate, well-structured, and consistently styled.
- Very specific: line 62
When repo and Outline disagree, the repo wins -- correct the Outline doc and flag it in the summary.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Outline Runbook Curator
Keep the StumpCloud Runbooks wiki accurate, well-structured, and consistently styled. The repo is the source of truth; Outline summarizes it for humans. This skill runs the same arc every time: discover (read-only) -> propose a plan -> get approval -> execute -> restyle -> verify.
The collection: StumpCloud / Runbooks (with OMGs as a sibling top-level section under the
same collection). Resolve its id with list_collections rather than hardcoding one, then
list_collection_documents on it to get the current tree.
House voice for wiki bodies: clear, operational, lightly dry -- but the wiki is reference material, so keep asides minimal. Facts and links are sacred. Blameless framing carries over from OMGs.
The arc
1. Discover (read-only -- never edit in this phase)
list_collection_documentsthe Runbooks collection for the live tree.fetcheach doc you intend to touch and note: current icon/color, ADR/SPEC citations, stale references (decommissioned hosts, services replaced by a successor, etc.), duplication with siblings, and thin/empty parent bodies.- Gather ground truth from the infrastructure repo -- ask the user for the checkout path if it
isn't already in context:
docs/adrs/(the ADRs the runbooks cite -- verify the claim against the actual file),decisions/,docs/anddocs-site/(narrative + user-facing guides), and the per-site inventory files +roles/for host/service facts. - For large trees, fan this out to subagents (one per section) that each return a concise per-doc report -- don't read 40 docs inline.
2. Propose a plan -- get approval before anything destructive
Summarize, in prose: what should merge/move/retire, every repo-vs-Outline discrepancy you'd correct, and any genuine judgment calls (with your recommended default). Do not execute merges, moves, or deletes until the user signs off. Low-risk, reversible moves can be batched once the overall plan is approved.
3. Execute
- Read before you edit. Prefer
editMode: "patch"withfindTextcopied verbatim, so rich formatting (highlights, comments, table widths) survives. Usereplaceonly for empty/one-line bodies you're fully rewriting. - Moves:
move_document(setparentDocumentId). Retire:delete_document(trashes, recoverable) -- prefer moving retired apps under a Deprecated node over deleting them. - When repo and Outline disagree, the repo wins -- correct the Outline doc and flag it in the
summary. Don't silently delete history; tombstone it (a dated
:::warning"retired/replaced by..." callout) instead.
4. Restyle
Color and icon every node per references/styling-conventions.md. Do this only after the
structure has settled.
5. Verify & report
list_collection_documents again to confirm the final tree. Deliver: what merged/moved/retired,
discrepancies corrected, and the final tree (with icons).
Core principles (full detail in references/)
- Hierarchy-native (
references/curation-principles.md): where sibling docs share a qualifier, that shared term is a missing parent -- create it and nest. Leaf titles shouldn't repeat what the breadcrumb already supplies. Parent nodes get a short orienting body that links their children; each child leads with its own distinguishing detail. - Repo wins (
references/curation-principles.md):docs/adrs/,decisions/, inventory, anddocs/are ground truth. Verify every ADR citation against the real file. - Styling (
references/styling-conventions.md): fixed per-domain colors on top-level nodes; each child gets its own distinct topical color (not a mechanical lightening of the parent); topical emoji per doc with no generic-icon reuse across siblings. OMG severity colors are reserved -- never restyle incident docs (see thestumpcloud-omgskill).
Related skills & references
stumpcloud-omg-- files individual incidents intoRunbooks / OMGs. Itsreferences/outline-formatting.mdis the canonical Outline markdown/callout cheat-sheet; reuse it rather than duplicating. Its severity color scheme is off-limits to this skill.references/styling-conventions.md-- the domain colors, intra-domain color rule, icon rules.references/curation-principles.md-- hierarchy-native structure, repo-wins reconciliation, the read-before-edit / patch discipline, and how to handle stale references.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

