From harness-engineering
Execute one micro-task from tasks.json using TDD. Use when user says "execute", "run next task", "implement next", "continue execution", or runs /execute. This is the fourth phase of Track 2. Enforces TDD gate: failing tests FIRST, then implementation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/harness-engineering:executorThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Execute exactly ONE micro-task from tasks.json. Enforces TDD (failing tests first)
Execute exactly ONE micro-task from tasks.json. Enforces TDD (failing tests first) and Chain of Verification (CoVe) before marking complete.
python3 ${CLAUDE_PLUGIN_ROOT}/skills/executor/scripts/select_next.py
Returns the next unblocked task (~100 tokens). If no tasks available, tell the user.
Read ONLY the files listed in the task's files array. Do not read unrelated files.
If the task has annotations, read those carefully — they contain human reviewer decisions.
Write failing tests FIRST:
Only after tests are validated:
After implementation, run this 4-step self-check:
python3 ${CLAUDE_PLUGIN_ROOT}/skills/executor/scripts/mark_complete.py \
<task_id> [--commit-sha <sha>]
This updates tasks.json and appends to claude-progress.txt.
Tell the user:
/execute for the next task."If tests don't pass after 2 implementation attempts:
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/progress.py append \
"2-Pass limit reached on <task_id>: <what was tried>"
Task: T001 — Add auth middleware to /api/users endpoint Process:
test_users_endpoint_requires_auth()Task: T003 — Implement token refresh logic Process:
test_token_refresh_extends_session()npx claudepluginhub emingenc/harness-engineering --plugin harness-engineeringImplements segmented tasks sequentially or by user specification, leveraging TDD commands for high-quality code. Integrates with Claude Code task system and documentation.
Orchestrates task execution with git worktree isolation, TDD implementation, validation loop, and merge for phrases like 'execute task N' or 'implement TASK-NNN'.
Executes tasks from TASK_N.md files or free-form descriptions, auto-generating missing scope, success criteria, and verification plans via /generate-tasks before implementation.