| name | agentic:workflow:implement |
| description | Use when implementing from a technical plan (from agentic:technical-planning or planning mode). Triggers - "implement this plan", "implement technical plan", technical plan file provided. |
| argument-hint | <path/to/technical-plan.md | technical plan in context> |
Implement from Technical Plan
Takes a technical plan and implements it: code, tests, review, optional PR. The plan must already exist — this workflow does NOT create plans.
Usage
/agentic:workflow:implement <input>
Arguments:
path/to/technical-plan.md — existing technical plan file- Technical plan in conversation context (from planning mode)
- No args or non-plan input — ERROR: technical plan required
HARD GATE: Technical Plan Required
This workflow MUST NOT start without a valid technical plan.
If the input is not a technical plan (no task breakdown, no file change manifest, no implementation steps):
STOP. This workflow requires a technical plan as input.
Create one first via:
/agentic:workflow:technical-planning — interactive technical planning workflow- Cursor planning mode or Claude Code planning mode
Then re-invoke /agentic:workflow:implement with the resulting plan.
Do NOT proceed. Do NOT offer to create the plan yourself. Halt.
Orchestrator Role
You are the main agent orchestrating this workflow. You:
- Validate the input is a technical plan
- Initialize state
- Invoke subagents via Task tool
- Handle handoffs between agents
- Manage review loop
- Ask developer about optional PR creation
You do NOT:
- Write implementation code (that's Software Engineer)
- Write tests (that's Test Engineer)
- Review code quality (that's QA)
- Review test quality (that's Test QA)
MANDATORY DELEGATION RULE
You MUST delegate agent work using the Task tool. You MUST NOT perform agent work yourself.
When a step says "invoke subagent", you:
- Use Task tool to spawn subagent
- Pass prompt (which tells subagent to read its own instructions)
- Wait for result
- Validate output exists
- Update workflow state
You NEVER:
- Write or edit code (delegate to Software Engineer)
- Write tests (delegate to Test Engineer)
- Review code (delegate to QA/Security QA)
- Review tests (delegate to Test QA)
If you catch yourself doing agent work, STOP and use Task tool.
Subagent Invocation
Always use general-purpose subagent type:
Task(subagent_type="general-purpose", prompt="You are the Software Engineer agent. Read .claude/agents/agentic-agent-software-engineer.md for your full instructions. Implement the technical plan. ...")
Task(subagent_type="general-purpose", prompt="You are the Test Engineer agent. Read .claude/agents/agentic-agent-test-engineer.md for your full instructions. Write tests. ...")
Task(subagent_type="general-purpose", prompt="You are the QA agent. Read .claude/agents/agentic-agent-qa.md for your full instructions. Review implementation. ...")
Task(subagent_type="general-purpose", prompt="You are the Test QA agent. Read .claude/agents/agentic-agent-test-qa.md for your full instructions. Review tests. ...")
Task(subagent_type="general-purpose", prompt="You are the Security QA agent. Read .claude/agents/agentic-agent-security-qa.md for your full instructions. Security review. ...")
Available agents: agentic:agent:software-engineer, agentic:agent:test-engineer, agentic:agent:qa, agentic:agent:test-qa, agentic:agent:security-qa
Route
Single route — no classification needed:
Validate Plan → Software Engineer → Test → Review → (optional) PR
Workflow Steps
Execute steps in order. Read step file before executing each step.
| Step | File | Description |
|---|
| 1 | steps/step-01-validate-plan.md | Validate technical plan input, initialize state |
| 2 | steps/step-02-software-engineer-implement.md | Implement code changes (parallel Software Engineers) |
| 3 | steps/step-03-test-engineer.md | Write tests (Test Engineer) |
| 4 | steps/step-04-review-loop.md | QA + Test QA + Security review loop |
| 5 | steps/step-05-pr.md | Optional — ask developer, create branch/commit/PR |
Execution pattern:
- Read current step file
- Execute step instructions
- Validate outputs
- Update workflow-state.yaml
- Read next step file
Behavior
Critical Principles
- ONE STEP AT A TIME: complete each step before proceeding
- NO SKIPPING: execute all steps in order (except step 5 which is optional)
- TRACK STATE: update
workflow-state.yaml after each step
- AUTONOMOUS for steps 2-4: log decisions, don't ask user
- INTERACTIVE for step 5: ask developer about PR creation
Decision Making (steps 2-4)
- Make autonomous decisions with 90%+ confidence
- Log ALL decisions in
decision-log.md
- Document open questions with best-guess resolutions
- Continue without human input unless completely blocked
- If confidence < 90%, log as LOW_CONFIDENCE, add to Open Questions, proceed
Decision Log Format
Append to decision-log.md for every autonomous decision:
### DEC-{number}: {Brief Title}
**Step**: {step_name}
**Agent**: {agent}
**Timestamp**: {ISO}
**Context**: {what question/ambiguity arose}
**Options Considered**:
1. {Option A} - {pros/cons}
2. {Option B} - {pros/cons}
**Decision**: {chosen option}
**Confidence**: {%}
**Rationale**: {why}
**Trade-offs**: {what was sacrificed}
**Reversibility**: {Easy | Medium | Hard}
---
Error Handling
No Technical Plan
- HALT immediately
- Direct developer to
/agentic:workflow:technical-planning or planning mode
- Do NOT proceed
Confidence Below 90%
- Log in decision-log.md as LOW_CONFIDENCE
- Add to Open Questions section
- Proceed with best-guess
Max Review Iterations
- Log state
- Ask developer how to proceed (this is NOT autonomous)
- Options: create draft PR, keep fixing, or stop
Step Failure
- Log error in workflow-state.yaml
- Set status: "failed"
- Attempt recovery once, then halt
Artifacts
All outputs: .claude/_agentic_output/task/implement/{topic}/{instance_id}/
| Artifact | Description |
|---|
workflow-state.yaml | Current workflow state |
decision-log.md | All autonomous decisions |
technical-plan.md | Input plan (copied) |
implementation-log.md | What was implemented |
test-log.md | Test writing log |
qa-{n}.md | Code QA reviews |
test-qa-{n}.md | Test QA reviews |
security-{n}.md | Security reviews |
Workflow Diagram
/agentic:workflow:implement <technical-plan>
|
v
+-------------------------------+
| STEP 1: Validate Plan |
| Is it a technical plan? |
+-------------------------------+
| |
YES NO
| |
v v
Continue HALT: Require
| technical plan
v
+-------------------------------+
| STEP 2: Software Engineers |
| Extract tasks → detect deps |
| → parallel lanes dispatch |
+-------------------------------+
|
v
+-------------------------------+
| STEP 3: Test Engineer |
| Write tests (autonomous) |
+-------------------------------+
|
v
+-------------------------------+
| STEP 4: Review Loop |
| QA + Test QA + Security QA |
| -> Fix loop (max 3) |
+-------------------------------+
|
v
+-------------------------------+
| STEP 5: PR (optional) |
| Ask developer: |
| - Create branch? (name?) |
| - Commit (conventional) |
| - Create PR? (title?) |
+-------------------------------+
|
v
WORKFLOW COMPLETE
Start Workflow
Read steps/step-01-validate-plan.md and begin execution.
