Evaluate-Loop Step 3: EXECUTE. Use this agent to implement tasks from a verified plan. Works through plan.md tasks sequentially, writes code, updates plan.md after every task, and commits at checkpoints. Uses TDD where applicable. Triggered by: 'execute plan', 'implement track', 'build feature', '/conductor implement' (execution phase). Only runs after plan has passed evaluation.
SkillScore64Maintained
Automatic
◆Coding pack
Last verified 8/23/2026
What it changes
Passed 10 of 16 test cases with the skill, and 4 of 16 without it.
Without skill27%
With skill64%
Checks it passed that the model alone missed
Marks the active task item with [~] syntax
Skips TDD test writing for CSS styling changes
Hands off to loop-execution-evaluator for step 4 evaluation
and 8 more in the full report
10 fixed4 passed either way6 still failing2 broke
Pass-rate lift
+36 pp
Tokens
+45%
Agent turns
0%
Response time
-62%
Model gemini-3.6-flashMeasured 8/23/2026v114 of 22 cases passed with skill · 16 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 10
Prompt
When starting Task 2 from plan.md, what markdown syntax change should be saved to plan.md before running any implementation commands?
What was checked, without and with the skill
Checked
Without skill
With skill
Marks the active task item with [~] syntax
Missed
Passed
case-05 · gemini-3.6-flash · 8/23/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md209 linessingle file
Loop Executor Agent — Step 3: EXECUTE
Implements the tasks defined in a verified plan.md. This agent writes code, creates files, and updates plan.md after every completed task.
Pre-Execution Checklist
Before writing any code:
read_file plan.md — find first [ ] task (skip all [x] tasks)
Confirm plan was evaluated (check for Plan Evaluation Report in plan or track metadata)
If no evaluation found → STOP → request Conductor run loop-plan-evaluator first
Execution Protocol
For Each Task
1. Mark task [~] in plan.md (in progress)
2. read_file acceptance criteria
3. Implement the task
4. Verify acceptance criteria met
5. Update plan.md immediately:
- Mark [x]
- Add commit SHA
- Add summary of what was done
6. Commit code changes
7. Move to next [ ] task
plan.md Update Format (MANDATORY after every task)
- [x] Task 3: Build signup form component <!-- abc1234 -->
- Created src/components/auth/signup-form.tsx
- Added email validation (regex), password min 8 chars
- Integrated with authApi.signUp() from mock API client
- Acceptance: ✅ Form renders, validates, submits