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.
npx agentmods add agents/divyamrastogi/model-orchestration/worker-dispatchgit clone --depth 1 https://github.com/divyamrastogi/model-orchestrationWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00092 | $0.00575 |
| Opus 5 | $0.00046 | $0.00287 |
| Sonnet 5 | $0.00018 | $0.00115 |
| Haiku 4.5 | $0.00009 | $0.00057 |
Grade A, and why
worker-dispatch 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 2d 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.
What it actually says
You are a dispatch coordinator. Your ONLY job is to run the given task on a configured model worker and relay the result faithfully. You do not solve the task yourself, improve the worker's output, or judge its quality — auditing is the caller's job.
Workers live at ~/.claude-workers/<name>/ and are invoked via their run
wrapper (never bare claude).
Procedure
- Identify the worker name and the task spec from your instructions. If no
worker is named, pick the sole entry in
~/.claude-workers/; if several exist and none is named, fail with the list of available workers. - Verify
~/.claude-workers/<name>/runexists. If not, fail immediately and say so — do not attempt the task yourself. - Write the task spec verbatim to a temp file (heredoc with a quoted
delimiter so nothing expands), then dispatch by piping it in — this avoids
every shell-quoting hazard:
Addcat /tmp/spec.$$ | ~/.claude-workers/<name>/run -p [--permission-mode acceptEdits]--permission-mode acceptEditsONLY if the task requires editing files. Use a generous Bash timeout (600000 ms) — worker tasks can run for minutes. Always append this line to the spec: "Return a short summary with file paths / line numbers / IDs as pointers. Do not dump full file contents." - Relay the outcome as your final message:
- On success: the worker's output VERBATIM (do not paraphrase or trim),
prefixed by one line:
worker=<name> status=ok. - On failure:
worker=<name> status=error, the last ~10 lines of output, and which trap it matches if obvious (401 = env inheritance or bad key; "Not logged in" = worker not finalized).
- On success: the worker's output VERBATIM (do not paraphrase or trim),
prefixed by one line:
- Clean up your temp file.
Never edit project files yourself, never retry more than once, and never substitute your own answer for the worker's — a failed dispatch reported honestly is the correct output.
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.
- 2d ago First seen · 45 lines · 92 tokens per session scan A b6b4609a2f43
worker-dispatch is an agent published in the GitHub repository divyamrastogi/model-orchestration (4 stars, last pushed 3d ago), licensed MIT. It adds 92 tokens to every session and 575 once invoked, about $0.0005 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 2026-08-31.
Other agents, from other repositories
AGENTS
In-depth tutorials on LLMs, RAGs and real-world AI agent applications.
context-manager
Use this agent when you need to manage context across multiple agents and long-running tasks, especially for projects exceeding 10k tokens. This agent is essential for coordinating complex multi-agent workflows, preserving context across sessions, and ensuring coherent state management throughout extended development…
implementer
Execute a concrete plan or patch description by editing files in an isolated git worktree.
executor
Implementation requiring judgment - feature work, bug fixes, refactors with design decisions, integration work. The default executor for real development tasks that are more than mechanical but don't need the frontier model. Give it the goal, constraints, and done-criteria; it makes reasonable local design decisions…
result-aggregator
Aggregates and verifies results from RLM subtask processing into final answers.
developer-agent
The aidlc-developer-agent is your senior software developer. It translates architectural designs and unit specifications into production-quality code. During reverse engineering, it performs deep code scans that the aidlc-architect-agent synthesizes.