Mola-maker/mstack / skills/m-dispatch
m-dispatch
VerifiedFan independent subtasks out to Claude subagents, DSH headless, Codex or Orca; isolate, merge, verify. Use for 分派/并行/开N个agents.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 10.7 KB |
| references/decisions.md | markdown | 1.1 KB |
| references/dsh-native.md | markdown | 4.9 KB |
| references/harness-commands.md | markdown | 8.7 KB |
| references/model-routing.md | markdown | 7.4 KB |
| references/subtask-prompt-template.md | markdown | 2.2 KB |
| scripts/cluster-diffs.mjs | script | 2.5 KB |
| scripts/cluster-diffs.test.mjs | script | 2.2 KB |
references/subtask-prompt-template.md
View on GitHub# Subtask prompt template (m-dispatch)
Use this template for every dispatch. Fill the placeholders, keep it self-contained, and do not paste accumulated history from earlier subtasks.
```
You are one of several parallel workers on `<project>`. You do not see the other
workers, and you must not touch anything outside your assigned files.
GOAL
<one concrete outcome, stated as a falsifiable predicate — e.g. "make these 3
tests pass" or "report every mismatch of X against Y">
WHY
<one or two lines: who needs this and what it unblocks, so the worker can
judge the calls this brief did not foresee>
SCOPE
- Files/dirs you may read: <paths>
- Files/dirs you may write: <paths, disjoint from every other worker>
- Constraints: <e.g. "do not change production code", "do not bump a version">
OUT OF SCOPE
<the nearby things a reasonable worker would also fix, and must leave alone>
CONTEXT
<only what this task needs: error text, file paths, interface signatures,
exact values. Point at files by path rather than pasting large diffs.>
VERIFY
<the exact command(s) that must pass, and the expected output — e.g. exit 0,
"N tests pass", "0 failures". Name the command, not "run the tests".>
RETURN
<the exact shape — changed files (or "no files changed") plus the VERIFY
command and the output you actually saw, or findings as
`file:line + problem + failure scenario`, at most N items. If this brief
names a SHA or a method, record both.>
Do not spawn further subagents. If you are blocked, report BLOCKED with the
exact missing input rather than guessing. Do not claim the selection rule;
the parent already wrote first pass, rank all, or best-of outside this prompt.
```
## Notes
- Leave out what the worker would read in the repo anyway (layout, conventions, test setup). Spend the words on what the repo cannot show: a decision, why the task matters, what done means this time.
- Put long context (a review package, a diff, a ledger) in a file and give the path, not the content.
- Evidence is the command, the output, and the changed files, plus any SHA or method the brief required. A RETURN that omits them is a gap: the parent re-dispatches that worker once, naming the missing field. A second miss stays a gap and does not count as a pass. Do not invent the missing output.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 469k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |