planr 0.0.1 → 1.1.16

package/docs/planr-spec/TASKS.md

@@ -0,0 +1,932 @@
+# Tasks
+
+These tasks are implementation-ready work items for coding agents. They intentionally avoid storypoints and timelines.
+
+## V1.1 Differentiation
+
+These tasks implement `V1_1_DIFFERENTIATION_CONTRACT.md`. They are Planr-owned product work and must not introduce public references to local comparison products.
+
+### TASK-V11-001: Establish V1.1 Gap Contract
+
+Goal:
+Create the executable V1.1 contract before coding.
+
+Context:
+Supports every REQ-V11-* requirement in `V1_1_DIFFERENTIATION_CONTRACT.md`.
+
+Requirements:
+- Define graph intelligence, pick recall, recovery, local browser review, scoped Git review, distribution, templates, docs, and E2E acceptance criteria.
+- Use Planr vocabulary only.
+- Keep the contract specific enough that coding agents can work each slice without hidden context.
+
+Files or areas likely involved:
+- `docs/planr-spec/V1_1_DIFFERENTIATION_CONTRACT.md`.
+- `docs/planr-spec/TASKS.md`.
+- `docs/planr-spec/README.md`.
+
+Acceptance criteria:
+- Contract exists with stable requirement ids.
+- Public docs contain no local comparison-product names.
+- Downstream implementation tasks can map directly to requirement ids.
+
+Tests:
+- Forbidden-reference scrub over README, docs, src, tests, examples, scripts, skills, `.github`, package metadata, and `.planr/project`.
+
+Dependencies:
+- V1 final validation gate.
+
+Do not do:
+- Do not implement feature code in this contract task.
+
+### TASK-V11-002: Implement Graph Intelligence
+
+Goal:
+Make Planr map analysis mathematically useful for agents and operators.
+
+Context:
+Supports REQ-V11-GRAPH-001 through REQ-V11-GRAPH-006.
+
+Requirements:
+- Compute critical path across blocking edges.
+- Compute transitive pressure and direct blocker counts.
+- Diagnose cycles before readiness and path output mislead agents.
+- Share one effect engine across close, cancel, insert, replan, and dependency previews.
+- Expose consistent results in CLI, MCP, HTTP, and JSON.
+
+Files or areas likely involved:
+- `src/app/repository.rs`.
+- `src/app/commands.rs`.
+- `src/app/http.rs`.
+- `src/app/mcp.rs`.
+- `tests/e2e.rs`.
+
+Acceptance criteria:
+- Branched and deep graph fixtures prove longest-path behavior.
+- Transitive bottleneck fixtures prove downstream impact ranking.
+- Cycle fixtures produce actionable diagnostics.
+
+Tests:
+- Focused unit tests for graph algorithms.
+- CLI/MCP/HTTP E2E coverage for critical path, pressure, lookahead, and previews.
+
+Dependencies:
+- TASK-V11-001.
+
+Do not do:
+- Do not treat priority sorting as critical-path analysis.
+
+### TASK-V11-003: Implement Automatic Pick Recall
+
+Goal:
+Make every pick return the context an agent needs to start safely.
+
+Context:
+Supports REQ-V11-RECALL-001 through REQ-V11-RECALL-005.
+
+Requirements:
+- Rank relevant contexts, plans, logs, and review summaries.
+- Include upstream handoffs, linked plan references, review/fix history, blockers, unlocks, possible file conflicts, and deeper-read commands.
+- Keep responses size-bounded and privacy-safe.
+- Share behavior across CLI, MCP, and HTTP pick routes.
+
+Files or areas likely involved:
+- `src/app/inspection.rs`.
+- `src/app/repository.rs`.
+- `src/app/commands.rs`.
+- `src/app/http.rs`.
+- `src/app/mcp.rs`.
+- `tests/e2e.rs`.
+
+Acceptance criteria:
+- Relevant historical decisions appear in pick output without a separate search.
+- Irrelevant contexts rank below relevant ones.
+- Prompt transcripts, source contents, secrets, and large artifacts are omitted by default.
+
+Tests:
+- Ranking fixture tests.
+- Pick output E2E tests across CLI, MCP, and HTTP.
+
+Dependencies:
+- TASK-V11-001.
+
+Do not do:
+- Do not dump all project context into every pick.
+
+### TASK-V11-004: Implement Recovery Automation
+
+Goal:
+Make interruption recovery deterministic and inspectable.
+
+Context:
+Supports REQ-V11-RECOVERY-001 through REQ-V11-RECOVERY-006.
+
+Requirements:
+- Add sweeper command/API with dry-run or preview-first behavior.
+- Add timeout and retry/backoff fields and transitions.
+- Reclaim stale picks only through explicit recovery behavior.
+- Surface preconditions and postconditions in pick, trace, review, and close preview.
+- Report all closure blockers in one gate report.
+
+Files or areas likely involved:
+- `src/storage/schema.rs`.
+- `src/model.rs`.
+- `src/app/repository.rs`.
+- `src/app/commands.rs`.
+- `src/app/http.rs`.
+- `src/app/mcp.rs`.
+- `tests/e2e.rs`.
+
+Acceptance criteria:
+- Interrupted work can be diagnosed, released, and safely re-picked.
+- Retryable failure returns to ready only after documented policy allows it.
+- Unsatisfied conditions block or warn exactly as documented.
+
+Tests:
+- Timeout and stale-pick tests.
+- Retry/backoff tests.
+- Close-preview condition tests.
+
+Dependencies:
+- TASK-V11-001.
+
+Do not do:
+- Do not silently mutate or reclaim active work by default.
+
+### TASK-V11-005: Build Local Browser Review Workspace
+
+Goal:
+Give reviewers a first-class local visual review workflow.
+
+Context:
+Supports REQ-V11-REVIEW-UI-001 through REQ-V11-REVIEW-UI-005.
+
+Requirements:
+- Serve a local review workspace from Planr.
+- Show plan/package review, plan diff, item detail, review evidence, annotations, and approve/request-changes actions.
+- Use existing review APIs to write annotations, artifacts, and follow-up work.
+- Keep the workspace local-first with no hosted dependency.
+
+Files or areas likely involved:
+- `src/app/http.rs`.
+- `src/app/review.rs`.
+- `src/app/surfaces.rs`.
+- `docs/`.
+- `tests/e2e.rs`.
+
+Acceptance criteria:
+- Browser smoke opens the workspace and exercises annotation flow.
+- Review feedback creates map-native fix/follow-up review work.
+- Clean review can write an artifact and close according to gate rules.
+
+Tests:
+- HTTP route tests.
+- Browser or HTML smoke tests.
+- Review artifact E2E tests.
+
+Dependencies:
+- TASK-V11-001.
+
+Do not do:
+- Do not require a cloud account or hosted paste service.
+
+### TASK-V11-006: Implement Scoped Git And Review Evidence
+
+Goal:
+Ground reviews in exact local source-change evidence.
+
+Context:
+Supports REQ-V11-GIT-001 through REQ-V11-GIT-005.
+
+Requirements:
+- Detect worktree, branch, commit, dirty state, and changed files.
+- Distinguish item evidence from unrelated dirty files.
+- Add file/line capable review evidence without storing full source contents by default.
+- Document local-only PR behavior or implement verified optional PR URL support.
+
+Files or areas likely involved:
+- `src/app/review.rs`.
+- `src/app/inspection.rs`.
+- `src/app/surfaces.rs`.
+- `tests/e2e.rs`.
+- `docs/OPERATING_MODEL.md`.
+
+Acceptance criteria:
+- Unrelated dirty files do not become agent-owned evidence.
+- Review artifacts list considered and excluded files.
+- File and line annotations survive export/import.
+
+Tests:
+- Git fixture tests.
+- Dirty-worktree review tests.
+
+Dependencies:
+- TASK-V11-001.
+
+Do not do:
+- Do not store full private source content by default.
+
+### TASK-V11-007: Harden Distribution And Client Setup
+
+Goal:
+Make Planr installable and configurable as a polished user product.
+
+Context:
+Supports REQ-V11-DIST-001 through REQ-V11-DIST-006.
+
+Requirements:
+- Keep release curl install as the primary user path.
+- Add Homebrew-ready path or explicit tap-publication condition.
+- Keep Cargo/source as maintainer workflow.
+- Add or document client setup for Codex, Claude Code, Cursor, and generic MCP.
+- Add prompt/config output for CLI, MCP, and HTTP workflows.
+- Document checksum verification.
+
+Files or areas likely involved:
+- `README.md`.
+- `docs/INSTALL.md`.
+- `docs/RELEASE.md`.
+- `src/app/commands.rs`.
+- `src/integrations.rs`.
+- `scripts/`.
+
+Acceptance criteria:
+- Fresh users can follow public release install docs.
+- Supported clients have documented setup.
+- Package dry-run contains only intentional files.
+
+Tests:
+- Install dry-run tests.
+- Package dry-run.
+- CLI prompt/config tests.
+
+Dependencies:
+- TASK-V11-001.
+
+Do not do:
+- Do not present npm or Cargo as the primary normal-user install path unless native binaries ship through that channel.
+
+### TASK-V11-008: Implement Templates And Review Packages
+
+Goal:
+Make proven Planr work reusable and portable.
+
+Context:
+Supports REQ-V11-TEMPLATE-001 through REQ-V11-TEMPLATE-005.
+
+Requirements:
+- Export/import graph items, links, plans, contexts, logs, review artifacts, and metadata.
+- Add package requirements metadata and preview-before-import.
+- Preserve review annotations, findings, artifacts, and file references.
+- Define or implement a local-first encrypted bundle strategy.
+
+Files or areas likely involved:
+- `src/app/inspection.rs`.
+- `src/app/repository.rs`.
+- `src/planpack.rs`.
+- `docs/IMPORT.md`.
+- `tests/e2e.rs`.
+
+Acceptance criteria:
+- Fresh project can import a proven decomposition and pick work.
+- Review annotations survive export/import.
+- Conflicts are reported before mutation.
+
+Tests:
+- Export/import E2E tests.
+- Package requirement and conflict tests.
+
+Dependencies:
+- TASK-V11-001.
+
+Do not do:
+- Do not rely on a hosted service for core package portability.
+
+### TASK-V11-009: Update Documentation And Public Copy
+
+Goal:
+Make docs match V1.1 behavior exactly.
+
+Context:
+Supports REQ-V11-DOCS-001 through REQ-V11-DOCS-005.
+
+Requirements:
+- Update README, CLI reference, MCP contract, operating model, task graph model, install docs, testing docs, and spec files.
+- Regenerate `docs/planr-spec.zip`.
+- Update fixtures that validate CLI/MCP/doc drift.
+- Scrub public repo paths for forbidden comparison-product names.
+
+Files or areas likely involved:
+- `README.md`.
+- `docs/`.
+- `docs/fixtures/mcp-contract.json`.
+- `docs/planr-spec.zip`.
+
+Acceptance criteria:
+- Docs, fixtures, CLI help, MCP tools, and tests agree.
+- Forbidden-reference scrub is clean.
+
+Tests:
+- Drift tests.
+- Forbidden-reference scrub.
+
+Dependencies:
+- TASK-V11-002 through TASK-V11-008.
+
+Do not do:
+- Do not document features that are not implemented or explicitly marked future scope.
+
+### TASK-V11-010: Run Full V1.1 E2E Verification
+
+Goal:
+Prove V1.1 works in-repo and as a fresh consumer install.
+
+Context:
+Uses the End-To-End Verification Contract in `V1_1_DIFFERENTIATION_CONTRACT.md`.
+
+Requirements:
+- Run all required repository checks.
+- Run fresh consumer E2E in `~/projects/planr-test`.
+- Include MCP, HTTP/SSE, browser review workspace, package export/import, and docs examples.
+- Convert any failure into fix and re-review tasks.
+
+Files or areas likely involved:
+- `tests/`.
+- `scripts/`.
+- `docs/TESTING.md`.
+- `~/projects/planr-test`.
+
+Acceptance criteria:
+- Every required command passes.
+- Consumer project evidence exists outside the source repo.
+- Final review has enough evidence to audit every V1.1 requirement.
+
+Tests:
+- Full command list from the V1.1 contract.
+
+Dependencies:
+- TASK-V11-009.
+
+Do not do:
+- Do not use narrow smoke tests as proof of broad V1.1 completion.
+
+## Foundation
+
+### TASK-FND-001: Create New Planr Product Skeleton
+
+Goal:
+Create a clean Planr codebase skeleton with owned command names, docs, and implementation boundaries.
+
+Context:
+Supports REQ-PROD-002 and ADR-001.
+
+Requirements:
+- Create package metadata for `planr`.
+- Create source-owned Planr files only through explicit Planr commands.
+- Add README sections for product direction, local-first data, and supported agents.
+
+Files or areas likely involved:
+- `Cargo.toml` or equivalent package manifest.
+- `README.md`.
+- `src/`.
+
+Acceptance criteria:
+- `planr --help` runs from source.
+- README no longer presents the product as Codex-only.
+- No unnecessary runtime dependency owns core coordination behavior.
+
+Tests:
+- Build command for chosen stack.
+- `planr --help`.
+
+Dependencies:
+- None.
+
+Do not do:
+- Do not copy unowned source code, docs, images, or command vocabulary.
+
+### TASK-FND-002: Implement CLI Command Skeleton
+
+Goal:
+Add the primary command groups and stable output conventions.
+
+Context:
+Supports CLIENT_IMPLEMENTATION_SPEC.md.
+
+Requirements:
+- Add command groups: project, plan, map, item, pick, log, review, close, context, search, doctor, mcp.
+- Add `--json`, no-color behavior, and stable error model.
+
+Files or areas likely involved:
+- `src/cli.rs`.
+- `src/app/commands.rs`.
+
+Acceptance criteria:
+- Each top-level command has help output.
+- Unknown commands and invalid arguments return machine-readable errors in JSON mode.
+
+Tests:
+- CLI golden help tests.
+- JSON error tests.
+
+Dependencies:
+- TASK-FND-001.
+
+Do not do:
+- Do not implement business logic in command parsing.
+
+### TASK-FND-003: Create `.planr` Project Pack Initializer
+
+Goal:
+Create a repo-local Markdown project pack with product, ownership, flows, state SSOT, constraints, and quality checks.
+
+Context:
+Creates the durable project context layer for Planr.
+
+Requirements:
+- `planr project init` creates `.planr/project/*.md`.
+- Existing files are not overwritten unless `--force`.
+- Starter text makes graph-vs-Markdown ownership explicit.
+
+Files or areas likely involved:
+- `src/planpack/`.
+- `.planr/` templates.
+
+Acceptance criteria:
+- Empty repo gets a complete `.planr` pack.
+- Existing `.planr` pack remains intact by default.
+
+Tests:
+- Init fixture tests.
+
+Dependencies:
+- TASK-FND-002.
+
+Do not do:
+- Do not add a second file-based graph SSOT beside SQLite.
+
+## Data And Graph
+
+### TASK-DATA-001: Implement SQLite Schema And Upgrades
+
+Goal:
+Create the local map database schema.
+
+Context:
+Supports API_AND_DATA_MODEL.md and ADR-002.
+
+Requirements:
+- Implement tables for projects, items, links, plans, source_links, contexts, runs, logs, artifacts, events.
+- Store schema version.
+- Enable WAL and foreign keys.
+
+Files or areas likely involved:
+- `src/storage/schema.rs`.
+- `src/storage/mod.rs`.
+
+Acceptance criteria:
+- Fresh database initializes.
+- Re-running init is idempotent.
+- Schema version is queryable.
+
+Tests:
+- Schema upgrade tests.
+- Fresh/open existing database tests.
+
+Dependencies:
+- TASK-FND-001.
+
+Do not do:
+- Do not store source file contents by default.
+
+### TASK-DATA-002: Implement Item State Machine And Atomic Picking
+
+Goal:
+Implement item lifecycle, link-based promotion, and concurrent-safe picks.
+
+Context:
+This is the core graph product.
+
+Requirements:
+- Centralize valid state transitions.
+- Promote ready items after link changes and item closure.
+- Implement atomic pick with pick token and automatically derived worker id.
+- Prevent parent closure while review/fix follow-ups remain open.
+
+Files or areas likely involved:
+- `src/core/state_machine.rs`.
+- `src/core/readiness.rs`.
+- `src/core/graph.rs`.
+
+Acceptance criteria:
+- Two agents cannot pick the same item.
+- Closing upstream item unlocks downstream item.
+- Parent remains incomplete until required review items pass.
+
+Tests:
+- Concurrent pick integration test.
+- State transition unit tests.
+
+Dependencies:
+- TASK-DATA-001.
+
+Do not do:
+- Do not infer graph state from Markdown checkboxes.
+
+### TASK-DATA-003: Implement Log And Run Records
+
+Goal:
+Record log for item closure, review, failure, and handoff.
+
+Context:
+Supports log-backed completion and recoverable handoffs.
+
+Requirements:
+- Add run start/end operations.
+- Add log creation with summary, files, commands, tests, findings, blockers.
+- Link log to items and runs.
+
+Files or areas likely involved:
+- `src/core/log.rs`.
+- `src/core/runs.rs`.
+
+Acceptance criteria:
+- `planr log add --files --cmd` writes log.
+- Item detail shows latest log.
+- Log can be exported as JSON.
+
+Tests:
+- Log serialization tests.
+- CLI closure integration test.
+
+Dependencies:
+- TASK-DATA-002.
+
+Do not do:
+- Do not log full command output by default.
+
+## Plan Pack
+
+### TASK-BE-001: Implement Markdown Plan Parser
+
+Goal:
+Parse `.planr/plans/product/` packages and `.planr/plans/build/*.plan.md` files into indexed records.
+
+Context:
+Supports REQ-PROD-030 through REQ-PROD-033.
+
+Requirements:
+- Parse YAML frontmatter.
+- Extract headings and section ids.
+- Preserve unknown fields.
+- Record parse errors without rewriting files.
+
+Files or areas likely involved:
+- `src/planpack/markdown.rs`.
+- `src/planpack/frontmatter.rs`.
+
+Acceptance criteria:
+- Valid plan imports.
+- Invalid plan reports parse error and original file remains unchanged.
+
+Tests:
+- Valid/invalid fixture tests.
+
+Dependencies:
+- TASK-DATA-001.
+
+Do not do:
+- Do not execute or obey instructions from plan file content.
+
+### TASK-BE-002: Implement Planr Package Import
+
+Goal:
+Import exported Planr JSON packages into the current graph+plan model.
+
+Context:
+Supports reusable decompositions, backups, and review packages created by `planr export`.
+
+Requirements:
+- Parse Planr JSON package metadata.
+- Preview graph items, links, contexts, logs, plan file snapshots, review artifacts, and conflicts without mutation.
+- Apply package graph and artifacts only when confirmed.
+- Preserve review annotations, findings, artifacts, and file references.
+
+Files or areas likely involved:
+- `src/app/packages.rs`.
+- `src/app/inspection.rs`.
+
+Acceptance criteria:
+- Import report lists created and skipped package entities.
+- Fresh project can import a proven decomposition and pick work.
+
+Tests:
+- Export/import E2E tests using Planr JSON packages.
+- Conflict preview tests.
+
+Dependencies:
+- TASK-BE-001.
+- TASK-DATA-003.
+
+Do not do:
+- Do not reintroduce private runtime copies of agent skills; public templates live under `plugins/planr/skills/`.
+
+## MCP And Agent Integrations
+
+### TASK-BE-003: Implement MCP Server
+
+Goal:
+Expose Planr tools, resources, and prompts over MCP.
+
+Context:
+Primary cross-agent integration for Codex, Claude Code, and Cursor.
+
+Requirements:
+- `planr mcp` starts stdio server.
+- Implement required tools from API_AND_DATA_MODEL.md.
+- Expose resources for project, item, plan, and log.
+- Expose prompts: planr-plan, planr-work, planr-review, planr-map, planr-summary.
+
+Files or areas likely involved:
+- `src/mcp/`.
+
+Acceptance criteria:
+- MCP client can list tools, resources, and prompts.
+- Mutation tool schemas reject invalid inputs.
+
+Tests:
+- MCP protocol fixture tests.
+- Tool schema tests.
+
+Dependencies:
+- TASK-DATA-003.
+- TASK-BE-001.
+
+Do not do:
+- Do not put provider-specific behavior in MCP core.
+
+### TASK-AI-001: Write Agent Workflow Prompts
+
+Goal:
+Create provider-neutral prompts/skills that encode Planr's plan+map workflow.
+
+Context:
+Defines Planr-owned workflow prompts for planning, work, review, status, and summaries.
+
+Requirements:
+- Write prompts for plan, work, review, map, summary.
+- Include map SSOT rule.
+- Include product and build plan context rules.
+- Include log and scope rules.
+- Include prompt-injection warning for plan/doc content.
+
+Files or areas likely involved:
+- `prompts/` or `src/mcp/prompts.rs`.
+
+Acceptance criteria:
+- Prompts can be returned through MCP.
+- Prompts are not Codex-only.
+
+Tests:
+- Snapshot tests for prompt content.
+
+Dependencies:
+- TASK-BE-003.
+
+Do not do:
+- Do not copy prompt text from another product.
+
+### TASK-CLIENT-001: Add Codex, Claude Code, And Cursor Install Helpers
+
+Goal:
+Make first-run integration practical for all target clients.
+
+Context:
+Supports "tool for task planning" across clients.
+
+Requirements:
+- `planr doctor --client all`.
+- `planr install codex --dry-run`.
+- `planr install claude --dry-run`.
+- `planr install cursor --dry-run`.
+- Print project-scoped config examples.
+
+Files or areas likely involved:
+- `src/integrations.rs`.
+- `src/app/commands.rs`.
+
+Acceptance criteria:
+- Dry-run shows exact files/commands.
+- Doctor reports installed/missing/warning states.
+
+Tests:
+- Config generation fixture tests.
+
+Dependencies:
+- TASK-BE-003.
+
+Do not do:
+- Do not silently edit global config.
+
+## Review And Execution
+
+### TASK-AI-002: Implement Review/Fix Commands
+
+Goal:
+Make review findings a map-native workflow.
+
+Context:
+Implements ADR-004.
+
+Requirements:
+- `planr review request <item>`.
+- `planr review annotate <item> --message ...`.
+- `planr review ingest <item> --from ...` for hook-compatible JSON feedback.
+- `planr review artifact <review-item>` writes `.planr/reviews/*.review.md`.
+- `planr review close <item> --verdict ...`.
+- Create fix and follow-up review work from findings, linked to the reviewed target.
+- Never auto-close or auto-approve work from ingested feedback alone.
+- Keep parent incomplete until review passes.
+
+Files or areas likely involved:
+- `src/app/review.rs`.
+- `src/app/commands.rs`.
+
+Acceptance criteria:
+- Review pass completes review item.
+- Review findings create fix/follow-up review chain.
+- Review close writes a review artifact and registers it in artifacts.
+- Parent remains blocked until clean review.
+
+Tests:
+- Review/fix loop integration tests.
+
+Dependencies:
+- TASK-DATA-002.
+- TASK-DATA-003.
+
+Do not do:
+- Do not use `failed` for ordinary review churn.
+
+### TASK-CLIENT-002: Add Optional Codex Runner Wrapper
+
+Goal:
+Provide a first-class Codex execution path without making Codex mandatory.
+
+Context:
+This is a product differentiator for Codex users.
+
+Requirements:
+- `planr codex run [--workers N]`.
+- Use `planr pick` to pick items.
+- Invoke `codex exec` with bounded item context.
+- Record runs and log.
+- Support `codex review` integration where configured.
+
+Files or areas likely involved:
+- `src/agents/codex.rs`.
+
+Acceptance criteria:
+- One item can be picked, run through Codex, and recorded.
+- Failed Codex run leaves item recoverable.
+
+Tests:
+- Mock Codex command tests.
+
+Dependencies:
+- TASK-DATA-003.
+- TASK-AI-001.
+
+Do not do:
+- Do not require Codex for Claude/Cursor users.
+
+## Security, Observability, Release
+
+### TASK-SEC-001: Implement Content-Safe Logging And Scrubbing
+
+Goal:
+Prevent accidental storage of secrets or private content in logs.
+
+Context:
+Supports SAFETY_PRIVACY_SECURITY.md.
+
+Requirements:
+- Redact likely API keys/tokens.
+- Keep prompt/response logging disabled by default.
+- Add `planr scrub`.
+
+Files or areas likely involved:
+- `src/security/`.
+- `src/observability/`.
+
+Acceptance criteria:
+- Secret-like strings are flagged in contexts/log.
+- Debug bundle excludes forbidden content by default.
+
+Tests:
+- Secret scrubbing tests.
+
+Dependencies:
+- TASK-DATA-003.
+
+Do not do:
+- Do not promise perfect secret detection.
+
+### TASK-ANA-001: Implement Doctor And Debug Bundle
+
+Goal:
+Provide local diagnostics for installation, database, MCP, and agent clients.
+
+Context:
+Supports production operability.
+
+Requirements:
+- `planr doctor`.
+- `planr debug bundle`.
+- Redacted output by default.
+
+Files or areas likely involved:
+- `src/doctor/`.
+- `src/observability/`.
+
+Acceptance criteria:
+- Doctor reports database, `.planr`, Git, MCP, Codex, Claude Code, Cursor.
+- Debug bundle is reviewable before sharing.
+
+Tests:
+- Doctor fixture tests.
+
+Dependencies:
+- TASK-CLIENT-001.
+- TASK-SEC-001.
+
+Do not do:
+- Do not include source files or plan bodies in debug bundles by default.
+
+### TASK-QA-001: Build Acceptance Test Suite
+
+Goal:
+Create the test suite described in QA_ACCEPTANCE_TESTS.md.
+
+Context:
+Prevents graph, package, and schema regressions.
+
+Requirements:
+- State machine tests.
+- Concurrent pick tests.
+- Plan parser tests.
+- MCP tests.
+- Schema upgrade tests.
+- Security/logging tests.
+
+Files or areas likely involved:
+- `tests/`.
+
+Acceptance criteria:
+- Acceptance suite runs locally.
+- CI can run it without external agent clients.
+
+Tests:
+- This task is itself the test suite.
+
+Dependencies:
+- TASK-BE-003.
+- TASK-BE-002.
+- TASK-SEC-001.
+
+Do not do:
+- Do not rely on live Codex/Claude/Cursor for core CI tests.
+
+### TASK-REL-001: Package And Release V1
+
+Goal:
+Ship Planr as installable local CLI.
+
+Context:
+Supports RELEASE_READINESS.md.
+
+Requirements:
+- Build release binaries.
+- Generate checksums.
+- Add install docs.
+- Add package import/export docs.
+- Add MCP integration docs.
+
+Files or areas likely involved:
+- `.github/workflows/`.
+- `docs/`.
+- release scripts.
+
+Acceptance criteria:
+- Fresh user can install and run `planr project init`.
+- Checksums are published.
+- Release notes list known limitations.
+
+Tests:
+- Packaging smoke test on macOS and Linux.
+
+Dependencies:
+- TASK-QA-001.
+- TASK-ANA-001.
+
+Do not do:
+- Do not ship install script that silently edits global agent config.