ZenUml/conf-app / .claude/skills/forge-ps-questionnaire
forge-ps-questionnaire
VerifiedFill, correct, submit, and audit an Atlassian Marketplace Privacy & Security questionnaire ("Edit responses") for the ZenUML apps (lite / full / diagramly / asyncapi), and verify the result via the Marketplace REST API. Use when updating or correcting a P&S declaration, when a customer security review exposes P&S answers that don't match actual data practices, when the same answers must be replicated across variants, or when auditing what each app currently declares. Encodes the field-name map, the native-setter-vs-keystroke rule (the sub-processor entity block needs real keystrokes or submit fails validation), the document.body screenshot workaround, per-variant deltas (asyncapi omits PlantUML), and REST verification (formStatus SUBMITTED vs APPROVED). Triggers on "privacy and security questionnaire", "P&S form", "Edit responses", "data security declaration", "sub-processor list", "CCPA/GDPR questionnaire", "marketplace privacy form", "audit P&S".
Forge Marketplace Privacy & Security questionnaire
Console form at marketplace.atlassian.com/manage/apps/<appId>/privacy-and-security →
Edit responses. There is no write REST API — edits happen only in the console
(browser). But the READ endpoint is public and is your verification backbone.
App map
| Variant | appId | appKey | sub-processors |
|---|---|---|---|
| lite | 1219422 | com.zenuml.confluence-addon-lite | 5 |
| full | 1218380 | com.zenuml.confluence-addon | 5 |
| diagramly | 1230599 | gptdock-confluence | 5 |
| asyncapi | 1225068 | my-api | 4 — no PlantUML (sequence macro stripped, so plantuml.com is unreachable; but AI Repair via the OpenAPI macro IS reachable → keep OpenAI/Anthropic) |
Workflow
- Audit current state (read-only, no browser):
scripts/verify-ps.sh <variant>— prints every field and flags!!deviations from the canonical target. Run this FIRST to see what actually needs changing. - Establish the answers. The authoritative template is an already-submitted
variant's live REST JSON (
scripts/verify-ps.sh <variant> --json), not a hand draft — the owner has already reviewed those answers. Apply per-variant deltas (asyncapi: 4 subs + "AsyncAPI/OpenAPI specification" EUD wording instead of "diagram source text"). - Fill in the console (browser). Delegate to a subagent and hand it
references/fill-mechanics.mdverbatim — that file holds the field-name map and the quirks that will otherwise cost hours (see "Non-negotiables" below). - Submit — click "Save and preview" → complete the flow. On the preview page, eyeball the fields the owner cares about (CCPA rows, sub-processors) before publishing.
- Verify it landed: re-run
scripts/verify-ps.sh <variant>.formStatusmust flip toSUBMITTED; if it still showsAPPROVEDwith old values, the submission did NOT land — say so, don't claim success.
Non-negotiables (each was a real, expensive failure)
- The sub-processor entity subform accepts ONLY real keystrokes (
browser_type,slowly:true). Programmatic values pass the DOM but fail "mandatory field" on submit. Everything else takes programmaticbrowser_evaluatewrites. See fill-mechanics.md. - One unsaved form at a time. Any navigation/reload/relay-drop discards the whole unsaved form. Do not open a second app's form (or run a PVT, or any other browser task) while an unsaved P&S form is in play. Finish → submit → then the next.
- Verify the precondition, not a verbal state. Before touching the browser, run
verify-ps.shto confirm whether the form is already submitted — don't act on "it's free" / "I submitted it" without theformStatuscheck. (Two forms' worth of work were lost to skipping this.) - Prefer fill-then-submit in one session over fill-then-stop-for-review. Review the decision once (on the first variant); for the rest, replicating an already-approved submission is mechanical — leaving it unsaved for review just exposes it to loss. Only stop-before-submit when the owner explicitly wants to eyeball that specific form.
- The listing JSON's
endUserLicenseAgreementUrlis dormant/stale — it keeps an old custom URL even when the app is on the Standard Agreement. Never judge the EULA from it; check the version's Links tab or the rendered listing page. (Mis-diagnosed 3 of 4 "EULA defects" from that field once.)
Canonical target (as of the 2026-07 security-review pass)
stores/process = Yes · residency = App does not support DR · retention max 90, custom
disallowed · GDPR controller = No, processor = Yes (+EUD types) · CCPA business = Not
applicable, service provider = Yes (+EUD types) · disk encryption = Yes · security policy
URL = https://app.zenuml.com/privacy-policy/privacy-policy.html · EEA transfer = Yes +
SCC mechanism · DPA = No (until one exists). verify-ps.sh flags anything off this.
Legal-judgment fields (don't silently flip; they're the owner's call, currently settled as above): GDPR controller No, CCPA business Not-applicable, CCPA service-provider Yes. The processor↔service-provider and controller↔business pairs should stay mirrored across all four variants. Mixpanel analytics is the one arguable "we're a controller" edge — flagged, deferred to counsel.
See references/fill-mechanics.md for the field-name map, native-setter/keystroke snippets, chip-list handling, and the screenshot workaround.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 470k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 470k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 470k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |