agentleFS
bookchaowalitbookchaowalit/platform-agent-skills/system-docpack/SKILL.md

system-docpack

Skill0 starsChanged 8 days agoApache-2.0View on GitHub

Create, complete, audit, and publish the per-system document pack (PRD, architecture, NFR, runbook, SRS, UX flows, test/UAT, RTM, business and delivery docs) and its PDFs for any Solo Empire portfolio system, client project, or our own apps. Use when asked whether a system's documentation is complete, to bring a system up to the EV3-style docpack, to add a new system to the portfolio docpack registry, or to rebuild docpack PDFs.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 3 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every portfolio system keeps one evidence-based document pack sized to its tier (L1 Lite, L2 Standard, L3 Full).

    line 8
  2. Very specific:

    Run task docpack:check (add -- --system <id> or -- --json) to see which systems and doc types are missing, malformed, or have stale PDFs.

    line 13
  3. Very specific:

    Read references/doc-standard.md for tiers, file names, front matter, required sections, and writing rules.

    line 15
  4. Very specific:

    Write each missing doc as NN-<doc_type>.md plus docpack.yaml.

    line 23
  5. Very specific:

    Commit only DOCPACK/ inside each nested repo.

    line 38
  6. Distinctive:

    No evidence for a fact → write [ต้องยืนยัน] / [สมมติฐาน] and an open question.

    line 44
  7. Distinctive:

    When asked whether a released system's documentation is complete, also check for a KPI plan with verified instrumentation (lifecycle …

    line 61

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

System Docpack

Every portfolio system keeps one evidence-based document pack sized to its tier (L1 Lite, L2 Standard, L3 Full). EV3 is the reference implementation.

Workflow

  1. Run task docpack:check (add -- --system <id> or -- --json) to see which systems and doc types are missing, malformed, or have stale PDFs.
  2. Read references/doc-standard.md for tiers, file names, front matter, required sections, and writing rules. Do not reload it per document.
  3. New system → add a row to repository-catalog/registries/docpack-coverage.yaml (id, name, kind, tier, pack). Choose the tier by risk: paid client acceptance → L3, public or small client/internal system → L2, internal tool or prospect → L1.
  4. Gather evidence from the owning boundary: code, configs, routes, migrations, notes, minutes, existing docs. Nested repositories are read-only.
  5. Write each missing doc as NN-<doc_type>.md plus docpack.yaml. Route content to the role skill that owns it: product-management (overview, prd, ux-flows), solution-architecture (architecture, nfr-security, runbook, trd-api-data), business-analysis (srs, rtm, business-case, business-flow, business-rules-data, decisions-glossary), qa-testing (test-uat), security-and-privacy (security and PDPA sections).
  6. Build PDFs with task docpack:build -- <system-id> [glob]. Packs marked pdf_builder: legacy (EV3) keep their own builder under the client's automation/docpack-pdf/.
  7. Personal Book Dev repos are discovered automatically and need a Product Card (L0). Collect facts with python3 infra/scripts/docpack/repo_facts.py, write DOCPACK/00-product-card.md with a verdict, then task docpack:cards:index. Promote an invest repo through repo_discovery.tier_overrides and write its L1+ docs in the same pack. Commit only DOCPACK/ inside each nested repo.
  8. Re-run task docpack:check; then task test:docpack if the checker or registry changed, and task docs:check before handoff.

Decision rules

  • No evidence for a fact → write [ต้องยืนยัน] / [สมมติฐาน] and an open question; never fill the gap with invented numbers, prices, or decisions.
  • Legacy pack with different file names → map types with doc_type_map in the registry or add a doc_type: front-matter line; do not rename files.
  • Client edition (audience: client) → the builder strips source sections, source columns, and repository paths; internal packs keep them.
  • Sending a PDF to a client, posting it, or uploading it anywhere is an external action → stop for explicit approval.
  • Personal names, contact details, secrets, raw client data → never in a pack.
  • The KPI plan is docpack type kpi-plan (KPI-PLAN, number 15), required for every Tier A/B product (registry product_tier_required; the checker does not enforce it yet). Write it from architecture-docs templates/sdlc/kpi-and-instrumentation-plan.md, usually at the repo's docs/measurement/, and map it into the pack. Launch plan, experiment log and outcome reviews stay outside the docpack in the project's SDLC workspace: link them from the overview (launch plan from the runbook, latest outcome review from the overview status).
  • When asked whether a released system's documentation is complete, also check for a KPI plan with verified instrumentation (lifecycle Measure) and an outcome review newer than its cadence whose decisions are backlog tasks (lifecycle Learn, architecture-docs docs/lifecycle/OUTCOME-REVIEW-TO-BACKLOG.md). Report a missing or overdue one as a gap and route it to product-management; do not invent figures to fill it.
  • For Tier A/B, report the Big Loop scorecard level from hub infra/scripts/platform/app_goal_scorecard.py --repo <name> (L3 needs the KPI plan, specs/ and a deploy workflow) next to the docpack result.
  • Also check the technical design detail documents listed in references/doc-standard.md (infrastructure and network, observability, capacity/performance/cost, failure modes, async integration, frontend, tech-debt register, landscape row, vendor exit plan). Work started on or after 2026-10-16 needs all of them (N/A + reason where a topic does not apply); older systems with no new work use the "Expected for" column. They are not docpack doc types, so task docpack:check will not flag them: report each missing one, or one without an N/A reason, as a gap and route it to solution-architecture. For the landscape, check the system's row in the portfolio file, not a per-system file.

Resources

  • references/doc-standard.md — tiers, layout, front matter, required sections, ID conventions, quality gate.
  • infra/scripts/utils/docpack_coverage.py — coverage checker (task docpack:check).
  • infra/scripts/docpack/make_pdfs.sh and build_html.py — PDF pipeline (task docpack:build); needs Playwright headless Chromium and Sarabun fonts in .runtime/fonts (or FONT_DIR).

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.