renatocaliari/execution-executor

Execution Executor

Autonomous plan execution orchestrator. Reads an approved plan from docs/, parses each scope by type, dispatches to the right executor, and consolidates results.

This skill is designed to run after the Plannotator gate approves the plan. It replaces manual step-by-step execution with a single autonomous orchestration pass.

---

Input

The skill operates on the approved plan document — the artifact persisted at docs/{YYYY-MM-DD}/{slug}/plans/spec-tech_{v}.md after the Plannotator gate passes.

Where {slug} is a short kebab-case identifier for the project (e.g. login-system, payment-refactor) and {v} is an auto-incremented version number.

The plan must contain scopes with type annotations:

• [TYPE] feature — implement new functionality
• [TYPE] optimization — improve a measurable metric (must include [METRIC])
• [TYPE] spike — research or prototype

If the plan has the optional "Execution routing" section (from product-planner), use it directly. Otherwise, infer routing from [TYPE] tags.

---

Role

You are an execution orchestrator — a senior engineering lead running a shift-left review of an approved plan. Your job is NOT to redesign or question the plan (that already happened in earlier phases). Your job is to execute every scope correctly, in dependency order, using the right tool for each type.

You have access to all pi tools and subagents. Use them.

---

Workflow

Step 1: Read and parse the plan

Read the approved plan file. Identify every scope and its type.

Example scope shape:

[SCOPE-1]
[TYPE] feature
Objective: Implement user login
Dependencies: None
DoD: User can log in with email/password
ACs: - Email and password fields validate
     - Successful login redirects to dashboard
     - Failed login shows error message

[SCOPE-2]
[TYPE] optimization
[METRIC] API P95 latency < 200ms (lower is better)
Objective: Optimize search endpoint
Dependencies: SCOPE-1
DoD: Search latency meets target

[SCOPE-3]
[TYPE] spike
Objective: Evaluate vector database options
Dependencies: None
DoD: Recommendation document with pros/cons

Build an execution plan respecting dependencies: scopes with no dependencies run first, dependent scopes wait.

Step 2: Report the execution plan

Before executing, present a clear execution plan to the user:

📋 Execution Plan for: {plan-name}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Phase 1 (parallel):
  ⏩ [SCOPE-1] Login — feature → worker
  ⏩ [SCOPE-3] Vector DB eval — spike → scout + researcher

Phase 2 (after SCOPE-1):
  ⏩ [SCOPE-2] Search optimization — optimization → autoresearch
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Ask the user:

Shall I proceed with autonomous execution? I'll report back when all scopes are complete.

If the user says yes, proceed autonomously. If no, ask what they'd like to adjust.

Step 3: Execute feature scopes

For each [TYPE] feature scope:

1. Spawn worker with the scope's DoD and ACs as the task:

   subagent({
     agent: "worker",
     task: `Implement [SCOPE-X]: {objective}
   DoD: {DoD}
   ACs: {acceptance criteria}
   Files in scope: {from plan or inferred}
   Constraints: tests must pass, no regressions`,
     context: "fork"
   })
   

2. After worker completes, run parallel-review:

   subagent({
     tasks: [
       { agent: "reviewer", task: "Review diff for correctness and regressions", output: false },
       { agent: "reviewer", task: "Review diff for simplicity and unnecessary complexity", output: false }
     ],
     concurrency: 2,
     context: "fresh"
   })
   

3. Apply feedback: synthesize reviewer findings and apply fixes worth doing now

4. If the scope involves UI/visual changes, run quality checks:

   • audit — accessibility (WCAG POUR), performance, theming, anti-patterns
   • critique — design review (heuristics, cognitive load, AI slop detection)

   Both require impeccable context. Run them after design context is established.

5. Mark scope as complete and move to the next

Step 4: Execute optimization scopes

For each [TYPE] optimization scope:

1. Check if autoresearch is already set up for this scope (look for existing autoresearch.md with matching metric)
2. If not set up, launch autoresearch setup:

   subagent({
     agent: "delegate",
     task: `Setup autoresearch for optimization scope [SCOPE-X]:
   Objective: {objective}
   Command: {infer from metric or use plan's suggested command}
   Metric: {metric name} ({unit}, {direction} is better)
   Files in scope: {from plan}
   Constraints: {from plan, e.g. tests must pass}
   Use /skill:autoresearch-create and configure the loop.
   Once configured, let it run autonomously.`,
     context: "fork"
   })
   

3. Set a stopping condition. Autoresearch loops forever by default. Set maxIterations in autoresearch.config.json or define a target:
   • If metric target is defined in the plan (e.g., API P95 latency < 200ms): stop when target is met
   • If no target: run for a reasonable number of iterations (10-20) or until improvements plateau
   • Use the autoresearch widget/dashboard to monitor progress
4. When autoresearch completes (target met or iterations exhausted), run parallel-review on the optimization changes
5. Mark scope as complete

Step 5: Execute spike scopes

For each [TYPE] spike scope:

1. Run parallel investigation:

   subagent({
     tasks: [
       { agent: "scout", task: `Investigate existing codebase for: {objective}. Find relevant files, patterns, and constraints.` },
       { agent: "researcher", task: `Research best practices and solutions for: {objective}. Provide concrete options with pros/cons.` }
     ],
     concurrency: 2,
     context: "fresh"
   })
   

2. Consolidate findings into a recommendation document at docs/{YYYY-MM-DD}/{slug}/plans/spikes/{scope-name}-decision.md (create the spikes/ subdirectory if needed)
3. If the spike reveals a code change is needed, optionally run parallel-review
4. Mark scope as complete

Step 6: Handle dependencies between scopes

• Scopes without dependencies can run in parallel (up to reasonable concurrency)
• If a scope depends on another, wait for it to complete first
• Use subagent with async: true and check status periodically for parallel phases
• After all scopes in a phase complete, proceed to the next phase

Step 7: Report results

After all scopes are executed, produce a consolidated report and save it:

Save to: docs/{YYYY-MM-DD}/{slug}/execution-report.md

📊 Execution Results: {plan-name}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
✅ [SCOPE-1] Login — feature — DONE (3 files changed, 2 reviews passed)
✅ [SCOPE-2] Search optimization — optimization — DONE (latency 180ms, target <200ms ✓)
✅ [SCOPE-3] Vector DB eval — spike — DONE (recommendation in docs/spikes/)

Timeline: {total duration}
Commits: {commit hashes for each scope}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Next steps:
- Review and merge branches
- Run full test suite

---

Parallel Execution Rules

• Independent scopes can run in parallel. Use subagent's async: true + concurrency to run multiple scopes simultaneously.
• Dependent scopes must wait. If SCOPE-2 depends on SCOPE-1, do not start SCOPE-2 until SCOPE-1 is complete and reviewed.
• Worktree isolation is available via worktree: true for scopes that might touch overlapping files. Use it when multiple feature scopes modify the same area.
• Reasonable concurrency: 2-3 parallel scopes maximum unless the plan explicitly allows more. Running too many in parallel increases the risk of conflicts.

---

Error Handling

• If a worker fails (crash, stuck, timeout): note it, log the error, and move to the next scope. Do not block the entire execution on one failure.
• If autoresearch crashes: check the log, fix if trivial, otherwise skip and note it.
• If a reviewer finds blocking issues: flag them but continue execution. Report them in the final summary. Do not halt the pipeline for review findings — the user will address them.
• If a spike is inconclusive: document what was learned and recommend next steps.

---

Execution Modes

This skill supports two modes, chosen at the start:

| Mode | Behavior | |------|----------| | Full autonomous | Execute all scopes without pausing. Report at the end. Best for overnight runs. | | Scope-by-scope | Execute one scope, present results, ask to proceed. Best for interactive oversight. |

The default is Full autonomous. Ask the user if they want scope-by-scope instead.

---

Workflow Position

This skill runs after the Plannotator gate approves the plan, replacing manual execution:

1. Shape Up Planning → spec-product.md (business rules, scope, risks)
2. [Optional] Interface Brainstorming → interfaces.md (wireframes, proposals)
3. Plan Critique → gap analysis on product spec + revision
4. Plannotator Gate → approves spec-product.md ← PRODUCT APPROVED
5. Tech Planning Sequencing → spec-tech.md (product context + tech scopes)
6. Execution Executor ← YOU ARE HERE
   ├── Read spec-tech.md (has product context + typed scopes)
   ├── Report execution plan → user confirms
   ├── Execute features → worker + parallel-review
   ├── Execute optimizations → autoresearch
   ├── Execute spikes → scout + researcher
   └── Report consolidated results to execution-report.md

---

How to invoke

Com pi-supervisor (recomendado)

O pi-supervisor é uma extensão que observa a conversa com um LLM separado (pode ser um modelo mais barato) e steering o agente de volta se ele desviar do objetivo. Use o slash command /supervise antes de começar:

/supervise Execute o plano aprovado em docs/{YYYY-MM-DD}/{slug}/plans/spec-tech_{v}.md
roteando scopes corretamente. Salve relatório em execution-report.md.

Depois que o supervisor confirmar (resposta "Supervision started"), prossiga:

/skill:execution-executor

O supervisor vai observar cada turno e, se o agente desviar do plano, injetar uma mensagem de steering para corrigir o curso.

Sem supervisor

/skill:execution-executor

From a parent agent (programmatic):

subagent({
  agent: "worker",
  task: "Execute the approved plan at docs/2026-05-12/login-system/plans/spec-tech_1.md using the execution-executor skill. Route each scope correctly and save the report at docs/2026-05-12/login-system/execution-report.md.",
  skills: ["execution-executor", "autoresearch-create"],
  context: "fork"
})

As a follow-up to product-planner:

After the product-planner produces an approved plan, the same agent (or a new one) can continue:

subagent({
  agent: "delegate",
  task: `The plan at docs/{YYYY-MM-DD}/{slug}/plans/spec-tech_{v}.md is approved. Execute it using execution-executor skill and save the report.`,
  skills: ["execution-executor"],
  context: "fork"
})

---

Interaction with Other Tools

| Tool/Skill | How this skill uses it | |------------|----------------------| | worker (pi-subagents) | Implements feature scopes | | reviewer (pi-subagents) | Reviews implementation diffs | | scout (pi-subagents) | Investigates codebase for spike scopes | | researcher (pi-subagents) | External research for spike scopes | | autoresearch-create | Sets up optimization experiment loops | | autoresearch.config.json | Controls max iterations for optimization scopes | | parallel-review (pi-subagents) | Runs adversarial review after implementation | | worktree (pi-subagents) | Isolates parallel feature work |

---

Output Expectations

Strong execution runs:

• Respect dependency order — no scope starts before its dependencies
• Use the right tool for each type — worker for features, autoresearch for optimization, scout for spikes
• Handle failures gracefully — one failed scope doesn't block the rest
• Produce a clear final report — what was done, what changed, what failed

Weak execution runs:

• Run everything sequentially when parallel is safe
• Use worker for optimization scopes (loses the autoresearch loop advantage)
• Ignore scope types and treat everything as implementation
• Block on minor failures or reviewer feedback