Work

Executes implementation tasks through agent teams using adaptive test-driven development.

1|Updated Apr 25, 2026
One-click install
npx skills add https://github.com/htxryan/claude-code-config-profiles --skill work-htxryan
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Work
Source: https://github.com/htxryan/claude-code-config-profiles/tree/main/.agents/skills/compound/work
Command: npx skills add https://github.com/htxryan/claude-code-config-profiles --skill work-htxryan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Coordinating multi-agent implementation work often leads to untracked tasks, missing acceptance criteria, and code merged without proper verification. This Skill orchestrates team-based TDD execution so every task is tracked, tested, and verified against its epic's requirements before closure. ## Core Features & Use Cases - Agent Team Delegation: Deploys an AgentTeam of test-writers and implementers that scale to the number of independent tasks, with the lead coordinating rather than writing code. - Verification Contract Enforcement: Reads each epic's Acceptance Criteria and Verification Contract, then requires the specified evidence (tests, lint, build, browser checks, CLI transcripts) before any task can be closed. - Beads Task Lifecycle: Tracks work via bd ready, bd update, and bd close, with a mandatory phase gate ensuring no work tasks remain open before review. - Use Case: Pick up a set of ready tasks from an epic, spin up parallel test-writer/implementer pairs, and close each task only after tests, lint, and the implementation reviewer approve. ## Quick Start Ask the agent to pick up the ready tasks from the current epic and implement them using the Work skill with test-driven development.

Frequently Asked Questions about Work

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run test-driven development with multiple AI agents?▼

Deploy an AgentTeam with paired test-writer and implementer agents, scaling the team to the number of independent tasks. The lead coordinates via SendMessage, reviews outputs, and never writes implementation code directly.

How do I track implementation tasks with beads?▼

Use bd ready to find available tasks, bd update <id> --status=in_progress when starting, and bd close <id> once all tests pass. The phase gate requires bd list --status=in_progress to return empty before review.

What is a Verification Contract in an epic?▼

A Verification Contract is a section in the parent epic listing required evidence such as test, lint, build, browser evidence, or command transcripts. Every listed item must be produced before the task can be closed; if missing, fall back to acceptance criteria and the test plus lint floor.

What happens if the implementation reviewer rejects the code?▼

If the implementation reviewer rejects the work, you must fix all reported issues, re-run the required evidence, and resubmit for approval. Closing a task without an APPROVED review is not permitted.

When should agents communicate during parallel implementation?▼

Agents working on overlapping areas or shared interfaces must coordinate via SendMessage before implementing. This prevents conflicts when multiple test-writer and implementer pairs work simultaneously on related code.