Skip to content
Agentmart
Submit a tool

dispatch

Hand approved work to one worker Bot.

Skill by douglasjarquin

IndexedMIT
Write a review

Chalk Labs

Need help deploying dispatch?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

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/douglasjarquin-dispatch

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

Hand approved work to one worker Bot. Use when the user asks for research, planning, investigation, or implementation.

Use cases: research

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 and code. Checked Nov 1, 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/dispatch

What's inside

Skill name
dispatch
Description
Hand approved work to one worker Bot. Use when the user asks for research, planning, investigation, or implementation.
License
MIT
Instructions
55 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Dispatch

## When to use it

The user gave an explicit approved request.
Research, planning, investigation, and implementation all go through this skill.
Do not use it to do that work in the coordinator chat.

## Required inputs and access

The Persist skill, so the brief exists at `/workspace/sum/tasks/<id>/brief.md`.
The worker procedure memory.
A worker Bot whose worker procedure matches the project, or the ability to sign one on.
Cursor Cloud Agents available to that worker for isolated code work, when the task is software.

## Sequence of work

1. Write the brief from the user's words.
   Do not investigate first to fill gaps.
   Name the repository, the outcome, the scope, and how to check the result.
2. Persist the brief under a new task id.
3. Reuse existing role workers (Marketing, Security, Personal, Operations, Square, and similar) for matching non-software work.
   Only sign a new Sum worker when none fits.
   Do not recreate Square/Cleaner or Atlas.
   Reuse a software worker whose worker procedure already matches, or sign one on when none fits.
   Write the worker procedure memory into that Bot's description.
   Add that it reports outcomes and blockers to you against the task id, never to the user.
   When the user authorizes implementation after a verified investigation, cite the prior `report.md` / task id in the brief and forbid redoing the investigation.
4. Before you message the worker, append an `intended` line to `/workspace/sum/action-log.md`.
   If that intent and target already has `intended` or `done` in the last six hours, append `skipped-duplicate` and do not message them unless the user explicitly asked for this dispatch again.
   A repeated native message is not that ask.
   Then message the worker with the task id and the brief path.
   Ask for the outcome back against that id.
   After the message, append `done` or `failed`.
5. Tell the user which worker started, what it will deliver, and the task id.
6. End the turn.

## How to validate the result

`/workspace/sum/tasks/<id>/brief.md` exists and matches the user's words.
The worker Bot description says it is not the coordinator.
The coordinator chat contains no research, planning, investigation, or implementation of the request.

## What to return

The task id, the worker Bot, and the outcome the user should expect.
Then stop.

## What requires approval

Work starts only from an explicit user instruction or an already-approved task.
An unapproved issue is not approval.
Do not start a second execution because a native message arrived twice.

From douglasjarquin/sum, MIT.

More from douglasjarquin/sum

Signals

Stars
0
Forks
0
Open issues
2
Last commit (this folder)
Oct 6, 2026 (26 days ago)

From GitHub (fetched Oct 24, 2026).

Details

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

More from douglasjarquin

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get dispatch running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying dispatch?

CallGet help
Cite this page

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