ce-work_skill

Overview

This skill executes engineering work plans efficiently while preserving quality and delivering complete features. It turns plan documents into a tracked sequence of actionable tasks, runs tests and checks, and coordinates commits and PR-ready artifacts. The focus is shipping finished, well-tested features that follow existing project patterns.

How this skill works

The skill reads a work document (plan, spec, or todo file), extracts implementation units, test scenarios, and verification criteria, and asks clarifying questions before beginning. It sets up an isolated branch or worktree, produces a prioritized task list, and selects an execution strategy (inline, serial subagents, or parallel subagents). For each task it implements code, writes tests, runs system-wide checks, and makes incremental commits until final validation and a PR-ready commit and operational validation plan are prepared.

When to use it

• You have a written plan or specification with implementation units and verification criteria.
• You need to convert high-level requirements into tracked, testable tasks and commits.
• When multiple tasks can be parallelized or when subagents help prevent context loss.
• Before making changes that touch callbacks, middleware, or cross-cutting concerns.
• When you must deliver a PR with tests, linting, and an operational validation plan.

Best practices

• Read the entire plan and resolve deferred questions before coding; get user approval to proceed.
• Work in an isolated branch or worktree; never commit to default branch without explicit permission.
• Use each unit’s Verification field as the primary done signal and write both unit and integration tests where applicable.
• Commit logical, test-passing units with clear conventional messages; avoid WIP commits when possible.
• Run system-wide checks: trace callbacks, test failure paths, and verify interfaces exposed through alternate entry points.
• Include a Post-Deploy Monitoring & Validation section in the PR describing logs, metrics, expected signals, and rollback triggers.

Example use cases

• Implement a new API endpoint from a feature spec, including models, services, and integration tests.
• Break a UI feature into component tasks, sync with Figma assets, and iterate until designs match.
• Resolve a bug that involves callbacks and error handling, adding integration tests to prove end-to-end behavior.
• Ship a refactor across multiple files using serial or parallel subagents to avoid context degradation.
• Prepare a production-ready PR with linting, full test suite, and an operational validation plan.