shinpr/recipe-fullstack-build

Execute Skill: llm-friendly-context before writing Agent prompts, handoffs, or generated artifacts. Execute Skill: subagents-orchestration-guide before making workflow decisions, invoking agents, or resolving findings.

Orchestrator Definition

Core Identity: "I am an orchestrator." (see subagents-orchestration-guide skill)

Local authority gate: Make this recipe's workflow decisions and validate each returned result directly; delegate semantic deliverable production to the named specialist.

Review Resolution Gate MANDATORY]: Resolve every actionable deliverable-review finding through subagents-orchestration-guide Review Resolution before correction or progression. Before the first finding disposition, read references/review-resolution.md from the loaded subagents-orchestration-guide skill.

Required Reference

MANDATORY: Read references/monorepo-flow.md from subagents-orchestration-guide skill BEFORE proceeding. Follow the Extended Task Cycle and Agent Routing defined there.

Execution Protocol

1. Invoke named specialists for deliverable production — pass deliverable paths between them and validate their results (see subagents-orchestration-guide "Orchestrator Execution Boundary")
2. Route agents by task filename pattern (see monorepo-flow.md reference):

• *-backend-task-* → task-executor + quality-fixer
• *-frontend-task-* → task-executor-frontend + quality-fixer-frontend

1. Follow the 4-step task cycle exactly: execute → branch on executor result → quality-fix → commit
2. Enter autonomous mode when user provides execution instruction with existing task files — this IS the batch approval
3. Scope: Complete when all tasks are committed or escalation occurs

CRITICAL: Run layer-appropriate quality-fixer(s) before every commit.

Work plan: $ARGUMENTS

Pre-execution Prerequisites

Work Plan Resolution

Before any task processing, locate the work plan. Resolution rule:

1. Use the work plan explicitly supplied in $ARGUMENTS when present.
2. Otherwise group layer-aware task files by the existing {plan-name}-{backend|frontend}-task-*.md naming contract and map each group to docs/plans/{plan-name}.md.
3. When task groups produce no candidate, use the only Work Plan under docs/plans/ when exactly one exists.
4. Select the sole candidate. When multiple candidates remain, present them for selection. When none exists, continue through the missing-prerequisite branch below.

Consumed Task Set

Compute the Consumed Task Set for this run — the exact files this recipe owns, executes, and later deletes. Use the same restricted pattern as Work Plan Resolution:

1. List task files in docs/plans/tasks/ matching the layer-aware patterns {plan-name}-backend-task-*.md and {plan-name}-frontend-task-*.md for the {plan-name} resolved by Work Plan Resolution. Single-layer tasks are excluded

Every subsequent reference to "task files" in this recipe — Task Generation Decision Flow, Task Execution Cycle iteration, and Final Cleanup — uses this set, not the unrestricted docs/plans/tasks/*.md glob.

Task Generation Decision Flow

Analyze the Consumed Task Set and determine the action required:

| State | Criteria | Next Action | |-------|----------|-------------| | Tasks exist | Consumed Task Set is non-empty | User's execution instruction serves as batch approval → Enter autonomous execution immediately | | No tasks + approved plan exists | Consumed Task Set is empty but the resolved work plan has batch approval | Run task-decomposer; the approval already authorizes mechanical task materialization | | No tasks + unapproved plan exists | Consumed Task Set is empty and the resolved work plan is not approved | Review it when needed, then present the plan approval gate before task materialization | | Neither exists + Design Doc exists | No plan, no Consumed Task Set, but docs/design/*.md exists | Invoke work-planner to create a work plan, then run document-reviewer (dev-workflows-fullstack:document-reviewer, doc_type: WorkPlan). Run Review Resolution through its correction re-review, escalation, and convergence transitions, using work-planner for rerouted corrections; then present the resolved plan for batch approval before task materialization | | Neither exists | No plan, no Consumed Task Set, no Design Doc | Report missing prerequisites to user and stop |

Task Materialization Phase (Conditional)

When the Consumed Task Set is empty:

1. Authorization Check

Use the normal Work Plan review and approval gate when batch approval is absent. Existing batch approval authorizes task materialization directly.

2. Task Materialization

Invoke task-decomposer using Agent tool:

• subagent_type: "dev-workflows-fullstack:task-decomposer"
• description: "Materialize work plan tasks"
• prompt: "Read work plan at docs/plans/plan-name].md and output individual single-commit task files in docs/plans/tasks/. Use layer-aware naming: {plan}-backend-task-{n}.md, {plan}-frontend-task-{n}.md from each Work Plan task's Executor lane."

3. Verify Generation

Recompute the Consumed Task Set using the same restricted pattern from the Consumed Task Set section above. When it remains empty, apply Specialist Result Acceptance: validate the invocation and returned artifacts, correct recoverable input or naming errors, and rerun. Stop for the user only when resolving the plan or intended layer boundary requires a user-owned decision.

Pre-execution Checklist

• ] Confirmed Consumed Task Set is non-empty (computed in the Consumed Task Set section above)
• ] Identified task execution order within the Consumed Task Set (dependencies)
• ] Environment check: Can I execute per-task commit cycle?
• If commit capability unavailable → Escalate before autonomous mode
• Other environments (tests, quality tools) → Quality agents retain proof limitations while the task cycle continues

Task Execution Cycle (Filename-Pattern-Based)

MANDATORY: For each task in the Consumed Task Set, route agents by task filename pattern from monorepo-flow.md reference.

Agent Routing Table

| Filename Pattern | Executor | Quality Fixer | |-----------------|----------|---------------| | *-backend-task-* | dev-workflows-fullstack:task-executor | dev-workflows-fullstack:quality-fixer | | *-frontend-task-* | dev-workflows-fullstack:task-executor-frontend | dev-workflows-fullstack:quality-fixer-frontend |

Task Execution (4-Step Cycle)

MANDATORY EXECUTION CYCLE: execute → branch on executor result → quality-fix → commit

Before the loop, register "Execute consumed task set", "Run post-implementation verification", "Clean up consumed task files", and "Report completion" once with TaskCreate; mark and advance the active phase with TaskUpdate.

For EACH task, YOU MUST:

1. EXECUTE: invoke Agent tool (subagent_type per routing table) → Record the current HEAD as diffBase, pass task_file: [path], and receive the structured response
2. BRANCH ON EXECUTOR RESULT:

• status: "escalation_needed" or "blocked" → Apply subagents-orchestration-guide Specialist Result Acceptance; escalate only a valid user-owned block
• requiresTestReview is true → Identify the changed integration/E2E test files in the current changes and invoke integration-test-reviewer with them as changedTestFiles, plus diffBase, taskFile, prompt-only claims, and mutationEvidence
• approved → Proceed to step 3
• blocked → Apply Specialist Result Acceptance
• needs_revision → Pass qualityIssues unchanged into the Review Resolution Gate; return to step 1 for rerouted corrections and derive convergence from correction re-review prior_feedback_reconciliation
• status: completed → Proceed to step 3

1. QUALITY-FIX: Invoke the layer-appropriate quality-fixer with task_file, upstream mutationEvidence, and qualityCommand when available (caller first, otherwise current task)

• stub_detected → Return to step 1 with the layer quality-fixer's incompleteImplementations array unchanged as the canonical incompleteImplementations field
• blocked → Apply Specialist Result Acceptance
• verification_incomplete → Retain the complete result for final retry and proceed to step 4
• approved → Proceed to step 4

1. COMMIT: Apply subagents-orchestration-guide Commit Boundary Check, then execute git commit after the layer-appropriate quality-fixer returns approved or verification_incomplete; append its verification trailers for the latter

Use each subagent's semantic result and repository evidence through Specialist Result Acceptance; canonical status fields provide the normal routing shortcut. Proceed to the next task after step 4 and retain any verification limitation with its status kept proof-limited.

Verify task files exist per Pre-execution Checklist, then enter autonomous execution mode. When requirement changes are detected during execution, escalate to the user with the change summary before continuing.

Post-Implementation Verification (After All Tasks Complete)

Before invoking post-implementation verifiers, apply subagents-orchestration-guide's retained verification limitation retry with each layer's quality-fixer. Continue with the verifiers after clearing or retaining each result; include only repeated limitations in the completion report.

Resolve all readable Design Docs from the Work Plan, or the Work Plan itself when none exist; missing input blocks verification.

Emit one code-verifier call per resolved document plus one security-reviewer call in one assistant message, then await all:

• code-verifier (subagent_type: "dev-workflows-fullstack:code-verifier") → verify the completed implementation against each resolved doc_type and single document_path
• security-reviewer (subagent_type: "dev-workflows-fullstack:security-reviewer") → review the completed implementation against the typed governingDocuments list

Apply subagents-orchestration-guide's Post-Implementation Verification status-routing and fix/re-run rules with the layer-appropriate executor and quality-fixer. Present the unified report; proceed to Final Cleanup after the complete verification set reaches Review Resolution convergence.

Final Cleanup

Before the completion report, delete the implementation task files this recipe consumed. Their work is committed; docs/plans/ is ephemeral working state and is not retained between recipe runs:

• Delete every file in the Consumed Task Set
• Preserve the work plan itself (docs/plans/{plan-name}.md) — the user decides whether to delete it after final review

If task-file deletion fails with a filesystem error, report the failure and continue to the completion report.

Completion Report Contract

Final report must include:

• Task materialization status
• Implemented task count, including backend/frontend counts
• Quality check result
• Verification limitations that remained after final retry
• Commit count
• Cleanup result
• Declined actionable findings with ID, governing reason, and evidence, when any occurred
• Escalation or blocking summary, if any