# Leftover: Step 5 (Decompose Tests) blocked on tracker auth

**Recorded**: 2026-05-14T20:51:00Z (Thursday)
**Blocker**: `user-atlassian-mcp` returns "Not connected" (verified via `getAccessibleAtlassianResources`).
**Type**: tracker availability — NOT a deferrable "non-user blocker"; the autodev tracker rule (`.cursor/rules/tracker.mdc` Tracker Availability Gate) requires explicit user decision (Retry / `tracker: local`).

## What is pending

Step 5 (Decompose Tests, tests-only mode) needs to run:
1. **Step 1t** — Test Infrastructure Bootstrap → creates `todo/[TRACKER-ID]_test_infrastructure.md` + matching Jira ticket
2. **Step 3** — Blackbox Test Task Decomposition → produces one task file per blackbox/perf/res/sec/res-lim scenario referenced in `_docs/02_document/tests/*.md`. Estimated 12–20 task files based on the current spec spread (FT-P-01…FT-P-18, FT-N-01…FT-N-08, NFT-PERF-01…NFT-PERF-04, NFT-RES-01…NFT-RES-08, NFT-SEC-01…NFT-SEC-13, NFT-RES-LIM-01…NFT-RES-LIM-04).
3. **Step 4** — Cross-Verification → produces `_docs/02_tasks/_dependencies_table.md` and verifies AC/restriction coverage.

Each task file must have a Jira ticket created inline (per `.cursor/skills/decompose/SKILL.md` Save Timing table) and then be renamed from numeric prefix to `AZ-<id>` prefix.

## Inputs ready

- `_docs/02_document/tests/environment.md` ✓
- `_docs/02_document/tests/test-data.md` ✓
- `_docs/02_document/tests/blackbox-tests.md` ✓
- `_docs/02_document/tests/performance-tests.md` ✓
- `_docs/02_document/tests/resilience-tests.md` ✓
- `_docs/02_document/tests/security-tests.md` ✓
- `_docs/02_document/tests/resource-limit-tests.md` (need to verify exists)
- `_docs/02_document/tests/traceability-matrix.md` ✓ (post-2026-05-14 drift Phase 2 re-issue, 97% in-scope coverage)
- `_docs/00_problem/{problem,restrictions,acceptance_criteria}.md` ✓ (post-drift-revision)

## Resolution paths

The next `/autodev` invocation MUST resolve one of:

- **(preferred) Retry auth**: User authenticates `user-atlassian-mcp` via Cursor's MCP UI; autodev then proceeds normally and creates AZ-prefixed task files with live Jira tickets.
- **`tracker: local` mode** (only with explicit user acceptance): tasks are written with numeric prefix + `Tracker: pending` header marker; state file's `tracker:` field is changed to `local`; a future invocation with a working Jira MCP runs a "Tracker Pending Sync" to back-fill tickets and rename the files.

## Step 4 deliverables (already applied — DO NOT redo)

- `Auth/JwtExtensions.cs` — JWKS refresh-interval optional config (C01)
- `Infrastructure/ConfigurationResolver.cs` — `ResolveOptionalPositiveIntOrThrow` helper (C01)
- `Dockerfile` + new `docker-entrypoint.sh` — runs `update-ca-certificates` at container start (C02)
- `docker-compose.test.yml` — passes 30s / 10s JWKS refresh intervals to `missions` (C01)
- `_docs/04_refactoring/01-testability-refactoring/{list-of-changes,deferred_to_refactor,testability_changes_summary}.md`

`dotnet build -c Release` clean (0 warnings, 0 errors). `ReadLints` clean on edited files.

## Replay procedure when Atlassian MCP is back

1. On next `/autodev`, the Bootstrap step (B1) reads this leftover, verifies MCP connectivity via `getAccessibleAtlassianResources`, and either:
   - **MCP works** → delete this leftover, autodev proceeds to Step 5 normally.
   - **MCP still down** → autodev surfaces the Choose A/B/C/D again (see `protocols.md`).
2. If the user chose `tracker: local` in the interim and tasks were created with numeric prefixes, the next "Tracker Pending Sync" walks `_docs/02_tasks/todo/*.md` looking for `Tracker: pending` headers, creates the matching Jira ticket per task, rewrites the header, and renames the file from `NN_xxx.md` to `AZ-<id>_xxx.md`.