| name | recipe-implement |
| description | Orchestrate the complete implementation lifecycle from requirements to deployment. |
Required Skills [LOAD BEFORE EXECUTION]
- [LOAD IF NOT ACTIVE]
subagents-orchestration-guide — agent coordination and workflow flows
- [LOAD IF NOT ACTIVE]
documentation-criteria — document creation rules and templates
- [LOAD IF NOT ACTIVE]
llm-friendly-context — clear prompts, handoffs, and generated artifacts
Spawn rule: every spawn_agent call uses fork_turns="none" so the subagent receives only the task message and explicitly provided context.
Full-Cycle Implementation
$ARGUMENTS
Orchestrator Definition
Core Identity: "I am not a worker. I am an orchestrator." (see subagents-orchestration-guide skill)
CRITICAL: MUST execute all steps, sub-agents, and stopping points defined in subagents-orchestration-guide skill flows.
ENFORCEMENT: Skipping any step or stopping point invalidates the entire workflow output.
CRITICAL Sub-agent Invocation Constraints
MANDATORY suffix for ALL sub-agent prompts:
[SYSTEM CONSTRAINT]
This agent operates within implement skill scope. Use orchestrator-provided rules only.
ENFORCEMENT: Sub-agent prompts missing the constraint suffix MUST be re-issued with the constraint appended.
Step 1: Requirement Analysis
Spawn requirement-analyzer agent to determine scale and affected layers.
[STOP — BLOCKING] Present requirement-analyzer output (scale, affectedLayers, scope) to user for confirmation. CANNOT proceed until user explicitly confirms.
When user responds to questions:
- If response matches any
scopeDependencies.question -> Check impact for scale change
- If scale changes -> Re-execute requirement-analyzer with updated context
- If
confidence: "confirmed" or no scale change -> Proceed to Step 2
Step 2: Layer-Based Workflow Routing
Based on requirement-analyzer output affectedLayers, route to the appropriate workflow:
| affectedLayers | Workflow | Reference |
|---|
["backend"] only | Backend Flow | subagents-orchestration-guide skill (Large/Medium/Small scale) |
["frontend"] only | Frontend Flow | See Frontend Flow below |
["backend", "frontend"] | Fullstack Flow | subagents-orchestration-guide references/monorepo-flow.md |
Backend Flow
Follow subagents-orchestration-guide skill Large/Medium/Small scale flow exactly. All steps, stopping points, and sub-agent sequencing are defined there.
Frontend Flow
STEP 1: Ask user if they have prototype code or UI references to provide.
STEP 2: Spawn ui-spec-designer agent → spawn document-reviewer agent.
[STOP — BLOCKING] Present UI Spec for user approval. CANNOT proceed until user explicitly confirms.
STEP 3: Spawn technical-designer-frontend agent → spawn document-reviewer agent → spawn design-sync agent.
[STOP — BLOCKING] Present Frontend Design Doc for user approval. CANNOT proceed until user explicitly confirms.
STEP 4: Spawn acceptance-test-generator agent → spawn work-planner agent → spawn document-reviewer agent with doc_type: WorkPlan.
[STOP — BLOCKING] Present Work Plan for user approval. CANNOT proceed until user explicitly confirms.
STEP 5: Run implementation readiness preflight.
Execute the Implementation Readiness Preflight Procedure from subagents-orchestration-guide for the approved work plan exact path. This means loading the work plan, evaluating R1-R5, resolving approved prep gaps through exact prep task files when needed, persisting the Readiness Report, and setting Implementation Readiness: ready or escalated. Apply the Implementation Readiness Marker Contract before entering autonomous execution.
STEP 6: Enter guided autonomous execution (see Autonomous Execution Mode below) using task-executor-frontend + quality-fixer-frontend agents.
Fullstack Flow
Follow subagents-orchestration-guide references/monorepo-flow.md for the complete cross-layer workflow, including:
- Separate Design Docs per layer
- design-sync for cross-layer consistency
- Vertical slicing in work-planner
- Layer-aware task execution routing
Autonomous Execution Mode
After user grants "batch approval for entire implementation phase", enter autonomous execution.
Implementation Readiness Gate
Before executing task files, read the associated work plan header and apply the Implementation Readiness Marker Contract from subagents-orchestration-guide.
Task Execution Quality Cycle (4-Step Cycle per Task)
Agent routing by task filename (for fullstack projects):
*-backend-task-* -> Spawn task-executor agent + quality-fixer agent
*-frontend-task-* -> Spawn task-executor-frontend agent + quality-fixer-frontend agent
*-task-* (no layer prefix) -> Spawn task-executor agent + quality-fixer agent (default)
Per-task cycle (complete each task before starting next):
- Spawn task-executor (or task-executor-frontend) agent: "Implement task [task-file-path]"
- Check task-executor response:
status: escalation_needed or blocked -> Escalate to user
requiresTestReview is true -> Spawn integration-test-reviewer agent
needs_revision -> Return to step 1 with requiredFixes
approved -> Proceed to step 3
- Otherwise -> Proceed to step 3
- Spawn quality-fixer (or quality-fixer-frontend) agent: "Quality check and fixes. Task file: [task-file-path]. The task file path above is also the
task_file input. Read its Quality Assurance Mechanisms section as supplementary quality-check hints. filesModified: [executor response filesModified]. Use these files as the stub-detection scope."
- Check quality-fixer response:
status: "stub_detected" -> Return to step 1 with stubFindings
status: "blocked" -> Escalate to user
status: "approved" -> Proceed to step 5
- git commit -> Execute on
status: "approved"
Post-Implementation Verification (After All Tasks Complete)
After all task cycles finish, collect all filesModified from every executor response (task-executor and task-executor-frontend, deduplicated), then run both verification agents before the completion report:
- Spawn code-verifier agent: "Verify implementation consistency against the Design Doc.
doc_type: design-doc. document_path: [path]. code_paths: [collected filesModified list]."
- Spawn security-reviewer agent: "Design Doc: [path]. Implementation files: [collected filesModified list]. Review security compliance."
- Consolidate results:
- code-verifier passes when
summary.status is consistent or mostly_consistent
- code-verifier fails when
summary.status is needs_review or inconsistent
- security-reviewer passes when
status is approved or approved_with_notes
- security-reviewer fails when
status is needs_revision
- security-reviewer
blocked -> Escalate to user
- If either verifier fails:
- Create a single fix task covering verifier discrepancies and security requiredFixes
- Spawn the layer-appropriate executor
- Spawn the layer-appropriate quality-fixer
- Re-run only the verifier(s) that failed
- Maximum retry count is 1 verification fix cycle; if any failed verifier still fails after re-run, escalate to the user
- If both verifiers pass -> Proceed to completion report
Test Information Communication
After acceptance-test-generator execution, when spawning work-planner, communicate:
- Generated integration test file path
- Generated fixture-e2e test file path or
null
- Generated service-integration-e2e test file path or
null
- E2E absence reason per lane when no E2E file is generated
- Note: integration tests are created with implementation; fixture-e2e runs alongside UI implementation; service-integration-e2e runs after all implementations when a service E2E file exists
Completion Criteria