Slash Command

/execute

Track 2 Phase 4: Execute one micro-task from tasks.json with TDD

npx claudepluginhub emingenc/harness-engineering --plugin harness-engineering

Allowed Tools

ReadGlobGrepEditWriteBashBash(python3 skills/executor/scripts/*.py:*)Bash(python3 scripts/*.py:*)

Prompt Preview

# /execute — Track 2 Execution

Follow the executor skill workflow exactly:

1. **Check context budget**:
   
   If `warning_level` is `warning` or `critical` → **STOP**: "Context budget at {utilization_percent}%. Recommend running /handoff before continuing."

2. **Select next task**:
   
   If no tasks available, tell the user. If all complete, congratulate.

3. **Read task context** — only files listed in the task's `files` array.

4. **TDD Gate**:
   - Write failing tests FIRST
   - Run tests — must FAIL
   - For M/L scope: show tests to user, wait for validation
   - For S scope: proce...

Command Content

Other plugins with /execute

/execute

1.4k

Executes specific tasks from Kiro specs with focused implementation for developer assistance. Accepts feature name and task description or number.

kiro-skill

/execute

316

Dispatches a single roadmap step to a specified agent for TDD execution, extracting context from roadmap.json, applying rigor profile, and updating execution-log.json.

/execute

242

Executes implementation plans systematically: loads plan, runs tasks in dependency order with checkpoints, validates against criteria, tracks progress, and reports metrics. Supports --task, --resume, --phase options.

attune

/execute

119

Executes development tasks in adaptive parallel waves with code review, quality gates, E2E checks, and task adaptation for spec-compliant complete implementations.

spectre

/execute

116

Executes validated plans using isolated agents for tasks, spec review, code review, and conflict resolution, with human checkpoints between batches. Supports --resume, --status, --dry-run, and other flags.

opus8 tools

claudikins-kernel

/execute

108

Executes an implementation plan from plans folders. Lists incomplete (DRAFT/APPROVED/IN_PROGRESS) plans for selection if no path provided, reviews, preps worktree for large tasks, runs autonomously.

7 tools

Stats

Stars0

Forks0

Last CommitMar 9, 2026

Actions

View Source View Plugin View on GitHub View README

# /execute — Track 2 Execution Follow the executor skill workflow exactly: 1. **Check context budget**: If `warning_level` is `warning` or `critical` → **STOP**: "Context budget at {utilization_percent}%. Recommend running /handoff before continuing." 2. **Select next task**: If no tasks available, tell the user. If all complete, congratulate. 3. **Read task context** — only files listed in the task's `files` array. 4. **TDD Gate**: - Write failing tests FIRST - Run tests — must FAIL - For M/L scope: show tests to user, wait for validation - For S scope: proce...

/execute — Track 2 Execution

Follow the executor skill workflow exactly:

Check context budget:

python3 ${CLAUDE_PLUGIN_ROOT}/scripts/context_tracker.py check

If warning_level is warning or critical → STOP: "Context budget at {utilization_percent}%. Recommend running /handoff before continuing."

Select next task:

python3 ${CLAUDE_PLUGIN_ROOT}/skills/executor/scripts/select_next.py

If no tasks available, tell the user. If all complete, congratulate.

Read task context — only files listed in the task's files array.

TDD Gate:

Write failing tests FIRST
Run tests — must FAIL
For M/L scope: show tests to user, wait for validation
For S scope: proceed automatically
Implement until tests pass

Chain of Verification (CoVe):

Does this handle edge cases?
Is this consistent with existing patterns?
Does this break existing tests?
Security concerns?
Fix any issues found.

Mark complete:

python3 ${CLAUDE_PLUGIN_ROOT}/skills/executor/scripts/mark_complete.py \
  <task_id> --commit-sha <sha>

Report what was done and suggest: "Context can be safely cleared. Run /execute for the next task."

2-Pass Rule

If tests don't pass after 2 attempts:

python3 ${CLAUDE_PLUGIN_ROOT}/scripts/progress.py append "2-Pass limit on <task_id>: <details>"

STOP and recommend context clear.

/execute — Track 2 Execution

Follow the executor skill workflow exactly:

Check context budget:
```
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/context_tracker.py check
```
If warning_level is warning or critical → STOP: "Context budget at {utilization_percent}%. Recommend running /handoff before continuing."
Select next task:
```
python3 ${CLAUDE_PLUGIN_ROOT}/skills/executor/scripts/select_next.py
```
If no tasks available, tell the user. If all complete, congratulate.
Read task context — only files listed in the task's files array.
TDD Gate:
- Write failing tests FIRST
- Run tests — must FAIL
- For M/L scope: show tests to user, wait for validation
- For S scope: proceed automatically
- Implement until tests pass
Chain of Verification (CoVe):
- Does this handle edge cases?
- Is this consistent with existing patterns?
- Does this break existing tests?
- Security concerns?
- Fix any issues found.

Mark complete:

python3 ${CLAUDE_PLUGIN_ROOT}/skills/executor/scripts/mark_complete.py \
  <task_id> --commit-sha <sha>

Report what was done and suggest: "Context can be safely cleared. Run /execute for the next task."

2-Pass Rule

If tests don't pass after 2 attempts:

python3 ${CLAUDE_PLUGIN_ROOT}/scripts/progress.py append "2-Pass limit on <task_id>: <details>"

STOP and recommend context clear.