docs-pipeline
Skill4 starsChanged 29 days agoMITView on GitHub
[PROTOCOL] Documentation backfill after an engagement changes code, architecture, deployment, or public contracts. Defines when to trigger, which doc files to update, and what the diff artefact must contain. Preloaded by lead agents (dev-lead, design-lead, marketing-lead). Pure reference — no triggers, not invoked by secretary.
- Reads credentials
In this repository48 other files it gives its agents.
- skills/dev-methodology/references/agents.mdAGENTS.md
- skills/skill-authoring/references/agents.mdAGENTS.md
- acceptance-protocolSkill
- agency-intakeSkill
- ai-visibility-methodologySkill
- banner-design-guideSkill
- benchmark-researchSkill
- brand-methodologySkill
- code-reviewingSkill
- code-writingSkill
- codex-bridgeSkill
- deploy-pipelineSkill
- design-assets-guideSkill
- design-system-methodologySkill
- design-task-decompositionSkill
- dev-methodologySkill
- documentation-writingSkill
- engagement-contractSkill
- engagement-protocolSkill
- feature-researchSkill
- human-voiceSkill
- infrastructure-setupSkill
- marketing-task-decompositionSkill
- post-deploy-qaSkill
- pre-deploy-qaSkill
- presentation-designSkill
- project-planningSkill
- prompt-engineeringSkill
- security-auditingSkill
- semantic-drift-methodologySkill
- seo-auditingSkill
- skill-authoringSkill
- skill-test-designSkill
- skill-testingSkill
- system-optimization-protocolSkill
- task-decompositionSkill
- tech-spec-planningSkill
- testing-methodologySkill
- ui-styling-guideSkill
- ui-ux-methodologySkill
- user-spec-planningSkill
- validation-pipelineSkill
- yandex-analytics-methodologySkill
- yandex-direct-guideSkill
- yandex-metrika-guideSkill
- yandex-search-guideSkill
- yandex-webmaster-guideSkill
- yandex-wordstat-guideSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 7 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 30
If engagement changed NONE of these, skip — but log "docs: no change" in engagement/docs-diff.md.
- Distinctive: line 71
Every proposed update landed in the right file.
- Distinctive: line 72
No doc bloat (writer didn't add redundant sections).
- Distinctive: line 74
No generic prose padding ("We value clean code.
- Distinctive: line 76
If issues: re-dispatch writer with specific fix list.
- Distinctive: line 81
Document quality against documentation-writing principles.
- Distinctive: line 82
No code blocks where prose suffices.
- Distinctive: line 84
No duplication across files.
- Very specific: line 124
Don't duplicate across files: If deployment.md lists the pipeline, patterns.md references it — doesn't repeat it.
≈ “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
docs-pipeline
Orchestrates documentation backfill as a cross-cutting pipeline run by leads before handoff. Keeps project-knowledge in sync with the code/artefacts actually produced.
Trigger — when to run
Lead MUST run docs-pipeline if the engagement changed ANY of:
| Domain | Trigger |
| dev | architecture (new service, layer, pattern), deployment (new target, env, secret), public API contract, build/CI topology, security posture, data model |
| design | brand voice/identity doc, design-system tokens/components, accessibility convention |
| marketing | brand voice/messaging, semantic core, reporting template, campaign pattern |
If engagement changed NONE of these, skip — but log "docs: no change" in engagement/docs-diff.md.
Project-knowledge layout
Canonical path: .claude/skills/project-knowledge/references/ in the project working directory.
| File | Owner content |
project.md | project overview, stack, environments, contacts |
architecture.md | system topology, services, layers, data flow |
patterns.md | code conventions, naming, folder layout, git workflow |
deployment.md | deploy targets, CI/CD pipeline, secrets table, rollback steps |
ux-guidelines.md | interaction patterns, voice, accessibility standards |
Design engagements may add: brand.md, design-system.md.
Marketing engagements may add: voice.md, semantic-core.md, campaigns-log.md.
Steps
1. Identify required updates
Lead diffs the engagement against existing docs:
- What decisions were made that are NOT in the docs yet?
- What code/pattern is now canonical that wasn't before?
- What operational step was added (new secret, new CI job, new rollback)?
List them in engagement/docs-diff.md under "Proposed updates".
2. Dispatch dev-technical-writer
Via Task tool. Pass:
- Absolute paths to the engagement artefacts (spec, PRs, decisions).
- Absolute path to
engagement/docs-diff.md(proposed updates list). - Absolute path to project-knowledge references directory.
- Engagement domain (dev / design / marketing).
Writer output: edits applied to references/*.md + a changelog section in docs-diff.md showing before/after per file.
3. Verify (lead)
Lead reads the diff. Checks:
- Every proposed update landed in the right file.
- No doc bloat (writer didn't add redundant sections).
- Code blocks contain only canonical snippets, not one-off engagement examples.
- No generic prose padding ("We value clean code. Our architecture is robust.").
If issues: re-dispatch writer with specific fix list. Max 2 writer re-dispatches per iteration.
4. documentation-reviewer pass
Dispatch documentation-reviewer via Task tool to check:
- Document quality against documentation-writing principles.
- No code blocks where prose suffices.
- Operational specifics present (paths, commands, secret names).
- No duplication across files.
Log verdict in engagement/validation-log.md under documentation-reviewer.
5. Record in docs-diff
engagement/docs-diff.md must contain by end of pipeline:
# Docs diff — {engagement-name}
## Files changed
- `references/architecture.md` — added section §3 Service X
- `references/deployment.md` — added secret NEW_API_KEY + rollback for Service X
- `references/patterns.md` — no change (no new pattern introduced)
## Review
- documentation-reviewer verdict: clean (or N findings + resolutions)
## Not yet written
- {if anything was deferred with reason}Skip rules
Allowed skip scenarios — log in docs-diff:
| Case | Justified because |
| Pure bugfix with no new pattern | nothing to document |
| Refactor within existing module, no API change | convention already captured |
| Copy tweak on existing landing | brand voice doc already lists voice rules |
Director will verify the skip justification matches engagement scope. Do NOT skip to save time if change is user-visible or operationally relevant.
Anti-patterns
- Don't treat docs as optional. Director rejects handoffs where architecture changed but
architecture.mddoesn't reflect it. - Don't dump code into docs. Project-knowledge is prose + tables + short canonical snippets, not a code dump.
- Don't write generic prose. "We value clean code" is doc bloat. Write specifics: "Folder
src/services/contains class-per-file, each class exports a single default." - Don't duplicate across files. If
deployment.mdlists the pipeline,patterns.mdreferences it — doesn't repeat it. - Don't skip
documentation-reviewer. Cheap check, catches rot.
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.

