work-on-task

work-on-task is a command for Claude Code from desplega-ai/agent-swarm. It costs 12 tokens per session (367 once invoked), scanned A, original, MIT.

A command for carrying out a task assigned through an agent swarm, where several coding agents share work. It retrieves the task, checks relevant saved notes, chooses a work method, records progress, and reports completion or failure.

In plain words
What is it for?
Use it to pick up or resume a research, development, or straightforward coding task in a shared agent workflow. Development tasks can be planned before implementation.
Why use it?
It gives an assigned agent a consistent way to understand its task and keep the lead agent informed. It also preserves useful lessons from earlier work.

Command for Claude Code

Written for Claude Code: argument-hint in frontmatter.

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/desplega-ai/agent-swarm/work-on-task
Clone the repo
git clone --depth 1 https://github.com/desplega-ai/agent-swarm

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for work-on-task

README.md
[![agentmods](https://agentmods.dev/badge/commands/desplega-ai/agent-swarm/work-on-task.svg)](https://agentmods.dev/commands/desplega-ai/agent-swarm/work-on-task)
Your own site
<a href="https://agentmods.dev/commands/desplega-ai/agent-swarm/work-on-task"><img src="https://agentmods.dev/badge/commands/desplega-ai/agent-swarm/work-on-task.svg" alt="Measured on agentmods" height="20"></a>
Per session 12 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 367 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00012 $0.00367
Opus 5 $0.00006 $0.00183
Sonnet 5 $0.00002 $0.00073
Haiku 4.5 $0.00001 $0.00037

Measured 7d ago against content hash 4d91139580ab, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

work-on-task scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugin/commands/work-on-task.md · 40 lines

What it actually says

Working on a Task

If no taskId is provided, call poll-task to get a new task.

Workflow

  1. Get task details: Call get-task-details with the taskId.

  2. Recall relevant memories: Use memory-search with the task description before starting any work. Past learnings, solutions, and gotchas are indexed here.

  3. Choose your approach based on the task type:

    • Research task → use /researching
    • Development task → use /planning first, then /implementing
    • Simple/direct task (no plan needed) → implement directly
  4. Work on it, calling store-progress at each meaningful milestone (not just start and end — the lead monitors this).

  5. Complete the task — see Completion below.

Completion

Call store-progress with:

  • Success: status: "completed" + output: "<what you did and the result>". Output should be specific enough for the lead to assess without re-reading your work.
  • Failure: status: "failed" + failureReason: "<what went wrong and what you tried>".

Then reply "DONE" to end the session.

Interruptions

If interrupted by the user, adapt to their instructions. When resuming, call /work-on-task <taskId> again to pick up where you left off.

When to escalate

If you're stuck after genuine effort (not just first failure), use /swarm-chat to ask the lead for help or clarification. Don't spin — escalate.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 7d ago First seen · 40 lines · 12 tokens per session scan A 4d91139580ab

Subscribe to this mod's changes

work-on-task is a command published in the GitHub repository desplega-ai/agent-swarm (743 stars, last pushed today), licensed MIT. It adds 12 tokens to every session and 367 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.