agentleFS
Sn4kyGitSn4kyGit/qa-agent-blueprint/.claude/skills/re-board/SKILL.md

re-board

Skill2 starsChanged 7 days agoApache-2.0View on GitHub

Bring an onboarded project up to the current blueprint schema without repeating onboarding — fills new template fields, seeds missing files, resolves open TODOs, refreshes bridge file + version stamp; every change individually confirmed. Use for "/re-board <KEY>", "re-onboard", "Projekt aktualisieren", after blueprint updates. Not for brand-new projects — /onboard-project.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    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. Very specific:

    After seeding, retro-learn stays the only writer, as always.

    line 42
  2. Distinctive:

    Run two read-only audits and add their findings to the review:

    line 50
  3. Distinctive:

    item by item: Every change is individually confirmed.

    line 73
  4. Very specific:

    If a Secrets source was set or changed to keychain: run the blueprint's resolve-secrets.sh (bash …

    line 77
  5. Very specific:

    After any confirmed PROJECT.md, PRODUCT.md or flow-map.md change (schema or hygiene), regenerate the bridge file …

    line 90
  6. Very specific:

    Update the Blueprint version stamp to the installed version once the schema delta is applied.

    line 95

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

/re-board <KEY> — delta onboarding for an existing project

Full onboarding (/onboard-project) costs an hour of interview. Re-asking it when the blueprint merely gained fields is how users learn to never update. This skill is the upgrade path: diff the project against the current template, offer only the delta, confirm each change individually, and re-stamp. Expect ~10 minutes for a typical project.

Steps

  1. Read projects/<KEY>/PROJECT.md, every projects/<KEY>/products/<PRODUCT>/PRODUCT.md, and their scoped learnings.md and flow-map.md files. If PROJECT.md does not exist, this is not a re-board — say so and route to /onboard-project.
  2. Build the schema delta list by comparing against the current template (projects/_template/; plugin mode: $CLAUDE_PLUGIN_ROOT/projects/_template/) and the installed blueprint version (.claude-plugin/plugin.json via the support link; plugin mode: claude plugin details qa-agent-blueprint or $CLAUDE_PLUGIN_ROOT/.claude-plugin/plugin.json):
    • Missing fields the template has and the project file lacks (e.g. Blueprint version, Secrets source, Framework, Readiness statuses, Readiness API/CLI, Build verification, Merge requirement, the whole Execution mode section — a project without it runs hybrid with the template defaults; offer the section with Mode as the one real question, sub-fields defaulted). Compare by field name, not by section text — a renamed heading is not a missing field.
    • Missing files: flow-map.md (project and each product). Offer a seed draft from recent evidence and learnings, confirmed before writing, under the same evidence gate as retro-learn: steps_executed yields only candidates and coarse paths; anchors require recon.json, verified spec steps, or explicit human confirmation and carry source:; neighbors require verified cross-flow evidence or explicit human confirmation and carry basis:, otherwise none — unconfirmed; last confirmed belongs only to a flow actually executed as that flow. Keep weaker entries under ## Unconfirmed candidates, outside confirmed Areas and bridge/retest scope. After seeding, retro-learn stays the only writer, as always.
    • Open TODO: entries from the original onboarding.
    • Version gap: installed version vs. the Blueprint version stamp (absent stamp = "pre-0.5.0"). Skim the CHANGELOG sections between the two (CHANGELOG.md via the support link; plugin mode: $CLAUDE_PLUGIN_ROOT/CHANGELOG.md) and note which entries actually concern this project — /whats-new does the full render, here one line per relevant change is enough.
  3. Run two read-only audits and add their findings to the review:
    • Hygiene audit across project/product context, learnings and flow maps: size and hot-context budget (the product template warns above 80 lines), duplicates, superseded or contradictory statements, volatile ticket states/repro narratives in durable files, and flow-map provenance (source:, neighbor basis:, unjustified anchors and confirmation stamps). Propose merge/condense edits and moves of rare or volatile knowledge to scoped learnings/evidence. Every consolidation is a separate confirmation item; never delete unresolved questions or either side of an unresolved contradiction. Budget findings warn and route knowledge to on-demand files — they never block a test run.
    • Harness delta report, using the established non-secret scan tooling rather than dumping config: Playwright --image-responses omit, --output-dir evidence/.mcp-out, .playwright-mcp.config.json console level warning, secrets wiring (classic --secrets or the qa-secrets.mjs mcp wrapper), optional raw-evidence read denies for .mcp-out and binary/image/trace formats (never blanket evidence/**), and legacy Atlassian URLs (?tools=all or v1/SSE). Label missing optional raw-evidence denies as hardening, not a contract failure. Report only; re-board never changes harness config. Point to the installer/update flow and session-start guidance for any remediation and restart.
  4. Present the schema and hygiene deltas as a numbered list, then walk them item by item. Every change is individually confirmed; nothing is bulk-applied. Present the harness report separately as read-only findings. A TODO the user still cannot answer stays a TODO — never fill a guess (same rule as onboarding).
  5. If a Secrets source was set or changed to keychain: run the blueprint's resolve-secrets.sh (bash $CLAUDE_PLUGIN_ROOT/.claude/hooks/resolve-secrets.sh (plugin mode) / bash .claude/hooks/resolve-secrets.sh (opencode/shared/integrate via symlink)). Then offer the assisted storage paths before manual commands: --import when the user already hand-filled values into .env.playwright-secrets (migrates declared keys into the keychain, values never surface), --capture for the rest (native macOS hidden-answer dialog per key — the user types into a GUI box, never into the chat). The manual one-time security add-generic-password command is the fallback for headless setups. Values never appear in chat, on any path.
  6. Refresh what depends on confirmed changes:
    • After any confirmed PROJECT.md, PRODUCT.md or flow-map.md change (schema or hygiene), regenerate the bridge file .agents/qa-project-context/<KEY>.md (and each affected product's) from the updated sources — with refreshed generated_from/source_sha, per the rules in /onboard-project. Include only confirmed flow-map Areas, never Unconfirmed candidates.
    • Update the Blueprint version stamp to the installed version once the schema delta is applied. Declined hygiene consolidations do not hold the stamp back. The project stamp records project-schema freshness only; it is not evidence that the harness or MCP config is current.
  7. Close with a compact summary: what was added/changed, what is still TODO:, remaining hygiene and read-only harness findings, and the one next action if anything is outstanding (usually: store keychain keys, fill .env.playwright-secrets, or run the installer/update guidance).

What this skill never does

Overwrite a filled field without asking. Create products/ (that is /onboard-product). Re-ask a question the file already answers. Fill a TODO with a guess. Print a secret value (the resolver's output discipline applies here too: key names only, ever).

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.