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 skills/jeonje/claude-plugins/ma-dispatchnpx skills add JeonJe/claude-plugins --skill ma-dispatchgit clone --depth 1 https://github.com/JeonJe/claude-pluginsWrote 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.
[](https://agentmods.dev/skills/jeonje/claude-plugins/ma-dispatch)<a href="https://agentmods.dev/skills/jeonje/claude-plugins/ma-dispatch"><img src="https://agentmods.dev/badge/skills/jeonje/claude-plugins/ma-dispatch.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00045 | $0.00777 |
| Opus 5 | $0.00023 | $0.00388 |
| Sonnet 5 | $0.00009 | $0.00155 |
| Haiku 4.5 | $0.00005 | $0.00078 |
Grade A, and why
ma-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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 32 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/ma-dispatch
Canonical form: /ma-dispatch <ISSUE_KEY> --project=<project> [--project=<project> ...].
/ma --auto without selection selects every approved project. A selected set must exactly be a subset of the approved project names; a path is evidence only, never authority.
Entry scope (the ma skill's scope gate (../ma/references/scope-gate.md)): scope must be exactly impl and implementation approval must be exactly approved. Existing implementation reconciliation is decided in scope-gate §5; do not reimplement validated work.
Before any worktree access, validate the key against the configured issueKeyPattern and validate project names, then read state. Check approved project/base/target equals each canonical state.worktrees[project]; reject unknown, duplicate, or unselected paths. The parent creates phaseResults["4"].targets[project] and allocates its absolute .claude/dispatch result/log paths (inside the worktree) before dispatch. Workers never write orchestrator state.
Target state is (every active worktree also records an absolute existing planPath inside that worktree):
{"baselineHead":"sha","status":"DONE_OK","attempts":[{"n":1,"mode":"plan-exec","status":"DONE_OK","startHead":"sha","endHead":"sha","result":"/abs/.claude/dispatch/run.result.json","log":"/abs/.claude/dispatch/run.log.md"}]}
Read references/delegation-discipline.md before composing any brief. Load prompts/executor-prelude.md, then the selected mode from prompts/templates.md. Use your workspace's executor agent (an agent role that implements with surgical changes and runs verification; run inline only if no such agent exists) with the target worktree as cwd. Its return is the short block plus the result JSON defined only in that prelude. plan-exec succeeds only with a new commit and exact final PLAN step coverage; review-fix/generic may say NO_OP explicitly.
On completion, verify each target's result yourself (a deterministic validator is not included in v1): the result JSON parses, matches the prelude schema, startHead/endHead are real commits in the worktree, planSteps exactly match the final PLAN headings, every verification is PASS for DONE_OK, and the log contains the embedded result comment. Only all-DONE_OK targets advance. Route DONE_FAILED, INVALID_RESULT, TIMEOUT, and SESSION_DIED separately. Retry only non-success targets, append attempts, and retain the original target baselineHead; never rerun a successful target. On a retry, set the executor's priorLog to the previous attempt's log so its deviations and PLAN corrections carry forward.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 4d ago First seen · 32 lines · 45 tokens per session scan A 0c5985c1eced
ma-dispatch is a skill published in the GitHub repository JeonJe/claude-plugins (3 stars, last pushed 1mo ago), licensed MIT. It adds 45 tokens to every session and 777 once invoked, about $0.0002 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 skills, from other repositories
kanban-dev-lane
Use when a Hermes Kanban worker wants to run an autonomous implementation delegation lane with a 3-tier fallback chain across claudy -> codex (--yolo) -> agyd (agy --dangerously-skip-permissions) -> Hermes direct on quota exhaustion.
process-backlog
Internal crew primitive, dispatched by /crew:run — do not invoke directly or select it for a user request; route backlog/catch-up requests to /crew:run, which decides the full workflow. (Function: find and advance stalled issues/PRs or ledger tasks, bounded; never merges.).
scrum
Scrum framework fundamentals, sprint goal writing, and agile ceremony facilitation. Use when the user asks to plan a sprint, write sprint goals, facilitate daily scrums, run sprint reviews or retrospectives, define scrum roles, manage the product backlog, or apply agile estimation techniques. Covers the three pillars…
survey
プロジェクト文脈を把握し .claude/context.md を作成する.新しいプロジェクトを始めるときに使う.context.md が既にある場合は /groundwork:theory を使う..
scrum-master
Agent team role for Scrum Master facilitation, coaching, and servant leadership. Use when the user asks to facilitate scrum events, coach a team on self-management, resolve impediments, run retrospectives, improve team processes, or address scrum anti-patterns. Owns the "how we work" - ensures the team applies scrum…
dev-plan
Full development planning pipeline. Orchestrates a team of agents (researcher, prd-writer, software-architect, project-planner, issue-estimator) to take a project idea from concept to sprint-ready issues.