Skip to content
Agentmart
Submit a tool

deepwork

Execute an already-approved, multi-step implementation plan with sustained focus, explicit ownership, and continuous verification.

Skill by Rafcin

IndexedMIT
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/rafcin-deepwork

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Execute an already-approved, multi-step implementation plan with sustained focus, explicit ownership, and continuous verification.

Use cases: agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/deepwork

What's inside

Skill name
deepwork
Description
Execute an already-approved, multi-step implementation plan with sustained focus, explicit ownership, and continuous verification.
License
MIT
Instructions
23 lines in SKILL.md · view on GitHub

Files (2)

  • SKILL.md
  • worktrees.md · references/

SKILL.md (first lines)

# Deepwork

## Entry conditions

Use only for an approved multi-step plan that the user asks to execute end-to-end. The plan and specification are authoritative. Confirm repository state, instructions, owned paths, and the first incomplete task; maintain a durable progress record for long work.

## Scope limit

Execute one task at a time with test-driven vertical slices and narrow attributable commits. Do not reset, stash, overwrite unrelated work, or widen a task because it is nearby. Delegate only under the declared OMCS route with exclusive write ownership.

Verify worker claims from the actual diff and fresh commands. A correction invalidates earlier verification and review.

## Exit evidence

Report completed plan steps, owned changes, focused and baseline evidence, current verification stage, and approval-gated proof still outstanding. Stop only for destructive or irreversible action, security-sensitive authorization, required external confirmation, or a plan defect with no grounded path.

For an explicitly isolated or parallel Git lane, read [Worktree lanes](references/worktrees.md) before creating, integrating, or removing a worktree.

## Next gate

Run `verification`, then the route-required fresh `code-review`; return a material plan defect to `context` or `plan`.

From Rafcin/oh-my-codex-slim, MIT.

More from Rafcin/oh-my-codex-slim

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Aug 26, 2026 (2 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from Rafcin

Cite this page

deepwork skill. Agent Mart. https://agentmart.dev/skills/rafcin-deepwork (last updated 2026-10-24)