Skip to content

dse120071750/m8m-harness-builder / skills/m8m/m8m-harness-builder/templates

__SKILL_NAME__

Verified

M8M 3.1 product skill whose closed native source compiles into a displayable, executable milestone workflow.

milestone/gem.md

View on GitHub
# __MID__ — __TITLE__

## Master prompt

You are responsible for completing **__TITLE__**. Start with this entire prompt,
then use the bound inputs and declared tools to produce the requested result.
This is the master prompt for milestone `__MID__`.

### Objective

__SUCCESS__

### Inputs and references

__INPUTS__

Interpret the supplied context semantically: notes, text, files, references, and
prior results may vary in shape. Do not require or validate an input schema or
insert a separate normalization milestone. Ask only for information actually
needed to perform the work. Tool calls still follow the tool's own API contract.

Use each input for its declared purpose. Follow the supplied specifications and
reference roles. Earlier milestones are available through these explicit bindings;
use their chosen outputs, including outputs from milestones before the immediate
predecessor. A mention in prose alone does not create an executable binding.

### Instructions

Carry out the complete task described above using the supplied domain instructions.
Preserve the specified identities, constraints, order, and reference relationships.
Resolve the inputs before calling the tools. If required input is missing, report
the missing input rather than inventing it. The numbered FlowSteps below identify
each internal action and its actual tools; they never replace this master prompt.

### Deliverable

Return the named outputs listed after the FlowSteps. Later milestones can bind
them using their declared references. Keep
this milestone's ID and output IDs stable when editing its prompt or tools.

Return the complete named output bundle in its declared format. File and media
outputs must point to actual files. Do not substitute a summary or a short
description for a requested complete prompt, document, image, or other artifact.
Intelligence may draft; it does not set `ok`, `branch`, or `cycle`.

Default to `loop: none`. Produce the named outputs and satisfy their schema;
do not add hashes, revisions, proof graphs, or automatic image reviews.
Run a separate review only when it is explicitly requested and configured.

The authored milestone `success` and output schema are the machine completion
contract. This master prompt supplies the full task instructions and agrees with
that contract. The runtime owns progress and completion.

- Milestone: `__MID__`
- Primary output kind: `__KIND__`
- Gem: `references/__MID__.md`
- Separate judge binding: `__WORKER__` (used only when `loop: judge`)
__JUDGE_LINE__

### Tool versus intelligence

__CLASSIFICATION__

Reuse existing functions, scripts, CLIs, or APIs through declared bindings.
Respect their existing approval and idempotency contracts. Do not discover other
runs or hand-write M8M chosen/control artifacts. Explicit packaging can impose
additional implementation restrictions; local coordination does not require
rewriting working tools.

Follow the numbered FlowSteps. If the preferred tool fails,
recover within the same complete master prompt.
The runtime admits the named result against this expectation before any
semantic judge. PASS commits only that admitted result to
`chosen-output.json`; missing or invalid declared members BLOCK first.
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: 472kUpdated Claude CodeCodexVerifiedInstall
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: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
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 availableStars: 322kUpdated Claude CodeCodexVerifiedInstall