# Work

> Use when the user explicitly invokes $work or asks for execution strategy for complex engineering work with unclear scope, dependent steps, or separable workstreams.

- Skill: `techdufus/work` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add techdufus/work`
- Raw SKILL.md: https://api.skillmd.com/api/skills/techdufus/work/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: techdufus (https://skillmd.com/u/techdufus)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/techdufus/work

---


# Work

Route complex work into the smallest execution shape that preserves evidence, focus, and
momentum. Do not duplicate narrower skills; let debugging, TDD, review, PR, or platform-specific
skills own their lanes.

## First Pass

- Outcome: name the concrete deliverable.
- Scope: identify files, systems, and non-goals.
- Evidence: find the source of truth before editing.
- Risk: decide what can break and how to check it.

## Mode

- `quick`: one obvious local step. Execute directly.
- `research`: user needs location, explanation, or confidence. Inspect source and answer.
- `structured`: dependent steps or multi-file edits. Use `update_plan`, then work sequentially.
- `parallel`: independent units with disjoint write scopes. Delegate only sidecar tasks when
  permitted; keep the blocking path local.
- `orchestrated`: large, staged work. Decompose into waves; wait only when blocked.
- `unclear`: ambiguity risks the outcome. Ask the smallest question.

## Noise Filter

- No plan for a one-step fix.
- No broad repo tour before the owning file or API is checked.
- No delegation for trivial or blocking work.
- No unrelated cleanup, refactor, or commentary.
- No completion claim without evidence.

## Closeout

Report what changed, what verified it, and any remaining risk or unverified edge.