ndisisnd/msg / .claude/skills/msg
msg
UnverifiedRoot menu for msg skills, plus harness modes. Bare `/msg` says the next step for each PRD in progress and offers to run it — use it when the user says "what's next", "where am I", "what should I run now". `--init` is the project setup — use it when the user says "initialise project", "bootstrap repo", "set up the framework", "start a new project", or asks to set up project structure in an empty repo; every msg skill also runs it by itself when `devkit/` is missing. Other modes: `--init-staging` (add a staging branch), `--update` (one pass that updates msg, pre-merge and merge settings, with one approval), `--gui` (local PRD board), `--doctor` (tally and triage the harness-incident ledger), `--aha` (sweep, triage, and compact the devkit/AHA.md learnings ledger), `--version` (print the installed release in one line), `--help` (guided skill picker).
msg
This file is a router. Every mode's protocol lives in a ref — none is written here. The Dispatch table below is the single statement of msg's mode surface — do not restate it elsewhere in this file.
Dispatch
Check the invocation before running any picker. First match wins.
| Invocation | Also triggers on (natural language) | Route |
|---|---|---|
/msg --init | "initialise project", "bootstrap repo", "set up the framework", "start a new project" | refs/protocol-init.md |
/msg --init-staging | "add a staging branch", "set up staging", "switch to a staged release flow" | refs/protocol-init-staging.md |
/msg --update | "check for msg updates", "reinitialise this project", "resync my init setup", "are there new init components", "update everything" | refs/protocol-update.md — msg's update, then pre-merge's and merge's --update, one change list, one approval |
/msg --gui, or the bare word gui | "open gui for PRDs", "show me the PRD board", "visualize my PRDs", "open kanban" | refs/protocol-gui.md — render directly, never call AskUserQuestion first |
/msg --doctor | "check the harness", "what keeps failing", "run the doctor", "triage the incident log" | refs/protocol-doctor.md — tallies and triages devkit/DOCTOR.md; never fixes |
/msg --aha | "prune the learnings", "sweep AHA", "trim the aha log", "triage the learnings ledger" | refs/protocol-aha.md — triages and compacts devkit/AHA.md; only write is the ledger itself, promotions stay recommendations |
/msg --version, or the bare word version | "which msg is this", "what version of msg do I have", "did the reinstall land", "is msg up to date" | refs/protocol-version.md |
/msg --help | — | refs/protocol-help.md |
/msg (no args) | "what's next", "where am I", "what should I run now" | refs/protocol-default.md — the next step per PRD in progress (script-msg-next.py), then run one or show the category/skill picker over the Skills table below |
Setup runs by itself. Every msg skill that finds no devkit/ runs /msg --init first
(its one interview), then continues the command the user asked for. /msg --init stays for
a manual run. --init takes no sub-flags; an unknown one is ignored with a one-line note.
Closing message. --init, --update, --init-staging, --doctor, and --aha runs end with the closing message
per ../shared/refs/closing-message.md — the last chat
output, after the protocol's own output. The pure-emission modes (bare /msg, --gui,
--version, --help) are exempt: their "Stop. Do not emit anything else." / render contracts stand unchanged.
Harness incidents. The same five modes log unexpected script failures, tool errors, retries,
and missed writes to devkit/DOCTOR.md per
../shared/refs/doctor-logging.md — logging never changes what
the run does next. --doctor is the reader, never a writer of incident rows.
Skills
| Category | Skill | Description |
|---|---|---|
| Planning | msg --init | Project setup — runs by itself on the first msg command when devkit/ is missing |
| Planning | intake | Capture + grade ideas/bugs into the INTAKE.md backlog (the front door); --update edits a captured row, --delete removes one, --health finds stale or wrong rows |
| Planning | plan-pm | Autonomous PRD writer — drafts from a graded intake row |
| Planning | plan-review | PRD contract certifier — seven consumer-bound checks; suggested after plan-pm and plan-em |
| Planning | plan-em | Engineering plan generator — certifies each wave, asks about the roster only when it is unusual |
| Build & Ship | eng | Plan or build engineering work from exec-table rows |
| Build & Ship | pre-merge | The quality gate — sync, mechanical, tests, regression, security/migration, PRD-consistency; fixes failures by itself (≤2 rounds); first run sets it up; stays local, ends "Next: /merge" |
| Build & Ship | merge | The ship gate — bare /merge does the next step; --staging (pre-merge pass check, push, PR, CI wait, merge, deploy, human test, sign-off), --production (confirmed release to main), --update (refresh settings); first run sets it up |
| Build & Ship | emulate | The local run lane — boots a simulator/emulator on the current branch and opens the window; a leaf, never part of the pipeline |
| Delivery | kermit | Conventional-commit formatter and changelog manager |
Footnote: This table is the canonical menu — it MUST list every user-facing skill in the msg workflow and any external skill the pipeline depends on (
kermit). When a skill is added, removed, or renamed, update this table and the--helprouting table (refs/protocol-help.md) in the same change. A skill absent from this table is unreachable through/msg.
End-to-end happy path
/msg --init → /intake → /plan-pm → (/plan-review) → /plan-em → (/plan-review)
↓
/eng --build
↓
/pre-merge (quality gate; then /merge)
↓
/merge (staging step: push, PR, CI, merge, deploy, human test)
↓
(human) /merge (production step: release to main)
(/plan-review) is a step the user chooses: plan-pm and plan-em suggest it, --review on makes them run it, and plan-em still certifies an unchecked PRD before engineering.
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: 467k | 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: 319k | 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: 319k | 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: 319k | Updated | Claude CodeCodex | Verified | Install |