dispatch

A command that sends a task to a chosen large-language-model command-line tool through a required planning and testing workflow. The first argument selects the model, and the remaining text describes the task.

In plain words
What is it for?
Use it to dispatch a task to a registered model such as Codex, Gemini, Qwen, Ollama, or another configured tool.
Why use it?
It makes delegated work follow a written specification with clear files, data, output, and restrictions. This reduces incomplete instructions and prevents the delegated tool from skipping required checks.

Command

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/rhan1/crewline/dispatch
Clone the repo
git clone --depth 1 https://github.com/rhan1/crewline
Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 653 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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 $0.00015 $0.00653
Opus 5 $0.00008 $0.00327
Sonnet 5 $0.00003 $0.00131
Haiku 4.5 $0.00002 $0.00065

Measured yesterday against content hash 6481959ac129, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dispatch scanned grade A with 1 finding 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 yesterday.

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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- UI component → dev server + browser/curl verification
commands/dispatch.md · 44 lines

How it starts

The opening of the file, as written. The whole thing — 44 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Run the dispatch workflow for the task below using the specified model. Do NOT shortcut the spec-writing or smoke-test steps.

The first word of $ARGUMENTS is the model ID (e.g. codex, gemini, qwen, ollama, llm). Everything after it is the task.

Task

$ARGUMENTS

Workflow

  1. Parse the model ID. Extract the first token of $ARGUMENTS as the model ID; the rest is the task description. If the model ID is blank or unrecognized, check ~/.claude/orchestrator-models.json for available models and ask the user which to use.

  2. Plan. Read the referenced files, pattern sources, and data endpoints yourself. If anything essential is missing (target path, output shape, data-source URL), ask before writing the spec — thin specs produce thin code.

  3. Write the spec to /tmp/dispatch-<model_id>-<short-task-name>-<unix-ts>.txt. Cover:

    • Exact target file path
    • 2–3 pattern files to mirror (style + error-handling conventions)
    • Data sources and expected fields
    • Output shape (JSON for APIs, component signature for UI)
    • Explicit "do not do" constraints — no npm/git/vercel/deploy, no tests/READMEs, no modifying other files, no network validation
  4. Dispatch via ~/.claude/scripts/llm-dispatch.sh <model_id> <spec-path> <short-task-name>. The wrapper runs the CLI, captures elapsed time + exit code to the model's last_file (path in ~/.claude/orchestrator-models.json), and tees the full log to ~/.claude/logs/llm-dispatch-<model_id>-<ts>.log.

  5. Smoke-test the output before claiming success:

    • API handler → mock req/res Node harness against live data
    • UI component → dev server + browser/curl verification
    • Script → run against real input
    • Analysis/summary → spot-check claims against source data
  6. Fix small bugs directly (< 10 lines). Re-dispatch only if the change is substantial. If the chosen model fails or rate-limits, re-dispatch to a different model ID — /dispatch codex <task> or /dispatch gemini <task>.

Read the full file on GitHub · 44 lines

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. yesterday First seen · 44 lines · 15 tokens per session scan A 6481959ac129

Subscribe to this mod's changes

dispatch is a command published in the GitHub repository rhan1/crewline (2 stars, last pushed 8d ago), licensed MIT. It adds 15 tokens to every session and 653 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.