po4yka/llm-wiki-skills/llm-wiki-team-rollout
Claude Code and agent skill by po4yka/llm-wiki-skills
- community
Checking that the SKILL.md file is still in the repository.
Plan and implement LLM-Wiki adoption for a team or company. Use for onboarding knowledge, bus factor, repo docs, decision records, PR-based agent writes, CODEOWNERS, permissions, governance, RACI/DACI, SLOs, review queues, team rituals, staffing, dashboards, security boundaries and operating models.
Install
npx vskill@latest install po4yka/llm-wiki-skills/llm-wiki-team-rolloutSKILL.md
LLM-Wiki Team Rollout
Goal
Introduce and operate LLM-Wiki in a team without creating an unreviewed official-looking slop layer or a hidden reliability burden.
Use references/docs/22-team-operating-model.md as the reference operating model for roles, RACI/DACI, governance, SLOs, rituals, staffing stages, dashboards, onboarding and GitOps controls.
When to use
- A team is adopting LLM-Wiki for the first time and needs an operating model, not just tooling.
- A pilot wiki is moving to early production and needs PR-based writes, CODEOWNERS and a review queue.
- Leadership asks for RACI/DACI, SLOs or a rollout plan before expanding to more domains.
- A team is deciding between read-only reports, PR-based writes, a draft wiki or direct writes.
- An existing rollout needs a governance or maturity-stage review (e.g. moving toward regulated/24x7).
Starter templates
Use these local templates when available:
references/templates/team-operating-model.yamlreferences/templates/team-raci-daci.yamlreferences/templates/team-review-workflow.yamlreferences/templates/team-slo-scorecard.yamlreferences/templates/team-onboarding-checklist.mdreferences/templates/CODEOWNERS.llm-wiki
Copy only the templates that fit the rollout stage, then replace placeholder roles, owners, SLOs and review-state gates.
Inputs
- Team size and workflows.
- Repository/docs platform.
- Existing docs, ADRs, onboarding material, issue/PR history, incident history and support channels.
- Security, privacy and permission constraints.
- Existing CODEOWNERS, branch protection and CI gates.
- Preferred write model: read-only, PR-based, direct writes, separate draft wiki.
- Current LLM-Wiki maturity: pilot, early production, business-critical, regulated/24x7.
- Desired outcome: adoption plan, operating model, RACI, templates, repo changes or governance review.
Procedure
1. Identify the organizational problem
Classify the primary use case:
- onboarding;
- bus factor / departing experts;
- agent context for coding workflows;
- decision provenance;
- incident/ops knowledge;
- product/research knowledge;
- customer/support knowledge;
- cross-team handoff;
- multi-domain knowledge governance;
- public/internal publishing;
- MCP/API/product interface support.
2. Choose maturity stage
| Stage | What is true | Suggested posture | |
|---|---|---|---|
| pilot | one domain, limited users, manual review acceptable. | one accountable pod, report-first metrics. | |
| early production | recurring ingest, CI evals, protected releases. | PR-based writes, named owners, weekly reviews. | |
| business-critical | multi-domain, API/MCP surface, publishing/export. | SLOs, incident playbooks, dedicated platform/security input. | |
| regulated or 24x7 | strict approvals, strong auditability, real on-call. | formal governance, support rotation, security sign-off. |
3. Define operating model
Recommended default:
Use roles:
| Role | Owns | |
|---|---|---|
| Product Owner | scope, value, priority, KPI targets, rollout. | |
| Technical Lead | architecture, schema, API/MCP, technical decisions. | |
| Knowledge Engineer | source model, wiki structure, taxonomy, provenance, page QA. | |
| Retrieval/Eval Engineer | retrieval quality, eval datasets, scorecards, regressions. | |
| Platform/SRE | CI/CD, observability, SLOs, runbooks, deployments, incidents. | |
| Security Engineer | threat model, data boundaries, approval policy, red-team. | |
| Domain Editor/SME | factual acceptance and domain judgment. | |
| Publishing/Docs Manager | information architecture, releases, exports, training. |
4. Choose write model
| Model | Use when | Risk | |
|---|---|---|---|
| read-only reports | early pilot or low trust | knowledge does not compound. | |
| PR-based writes | most teams | review queue required. | |
| separate draft wiki | high automation or sensitive domains | promotion workflow needed. | |
| direct writes | tiny trusted teams only | silent corruption. |
Default to PR/proposal-based writes for teams:
5. Define ownership and RACI
Map:
- who owns each wiki domain/path;
- who approves synthesis pages;
- who owns taxonomy/schema changes;
- who handles lint and eval failures;
- who owns ingestion profiles;
- who approves MCP/API/export changes;
- who handles incident response;
- which domains are sensitive or restricted;
- what agents may read, propose and publish.
Use RACI for recurring execution and DACI for cross-team decisions.
Minimum RACI fields:
6. Add governance layers
Ensure the rollout covers:
| Layer | Mechanism | |
|---|---|---|
| domain ownership | path owners, CODEOWNERS, SMEs. | |
| technical ownership | architect, ADRs, schemas, MCP/OpenAPI contracts. | |
| quality governance | evals, scorecards, lint, review backlog. | |
| security governance | threat model, model policy, redaction, incident playbook. | |
| publishing governance | export profiles, manifests, release approvals. | |
| operational governance | SLOs, runbooks, alerts, error budget policy. |
7. Start with high-value domains
Good first domains:
- onboarding map;
- architecture overview;
- ADR/decision history;
- repo conventions;
- incident postmortems;
- product terminology;
- recurring support questions;
- policy/runbook pages with clear owners.
Avoid migrating everything at once.
8. Define metrics and SLOs
Track:
- retrieval hit rate;
- citation coverage;
- unsupported-claim rate;
- review backlog count;
- median review age;
- stale verified pages;
- source/index freshness lag;
- agent mistakes caused by missing context;
- PR cycle time for documentation updates;
- docs read/write ratio;
- onboarding questions answered from the wiki;
- public/agent export failures;
- incident count and MTTR;
- team toil percentage.
Example gates:
9. Define operating rituals
Recommended cadence:
| Cadence | Ritual | |
|---|---|---|
| daily async | ingest/retrieval/security triage. | |
| weekly | retrieval/eval review. | |
| weekly | review queue grooming. | |
| weekly/biweekly | release/export review. | |
| biweekly | architecture/security review. | |
| monthly | team health and toil review. | |
| monthly | scorecard review. | |
| quarterly | operating model review. | |
| per incident | incident command and postmortem. |
10. Add GitOps controls
Recommend:
- CODEOWNERS for
wiki/,raw/manifests/,skills/,templates/,.github/,api/,mcp/,exports/profiles/,evals/; - branch protection or rulesets;
- required PR reviews;
- required checks by PR type;
- protected environments for publishing/deployment;
- minimal workflow permissions;
- artifact manifests and scorecards.
11. Create rollout plan
Use phases:
- team charter, working agreements, owners, initial RACI;
- read-only inventory;
- draft wiki for one domain;
- PR-based updates and review states;
- weekly lint/eval/security review;
- onboarding/query usage;
- export/MCP/API only after gates exist;
- expand to more domains;
- operating scorecard and maturity review.
Output
Safety gates
- Do not recommend direct writes for teams without a named reviewer and rollback path.
- Do not index restricted material into broadly readable wiki pages.
- Do not blur draft and verified knowledge.
- Do not let lint/eval agents silently resolve truth conflicts.
- Do not recommend public export,
llms-full.txt, MCP/API or agent-bundle rollout before redaction, eval and approval gates exist. - Do not recommend 24x7 support without a sustainable on-call staffing plan.
Security
- Security scanPassed, 100 of 100Pattern checks on the skill's files.
- Model reviewNot runA model review is scheduled for the next rescan.
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/llm-wiki-team-rollout/SKILL.md
Popularity
- Repository stars
- 3
- 7-day trend score
- 7
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/po4yka/llm-wiki-skills/llm-wiki-team-rollout)