romerma

26 mods across 1 repository, 1 stars between them.

mstack

01

romerma/mstack

Plugin Claude Code

The mstack plugin: rigorous agent workflows with gates that actually gate.

1 8d ago A tokens not measured original MIT

mstack

02

romerma/mstack

Plugin Claude Code

Rigorous, verifiable agent workflows for Claude Code. A router over playbooks, an opt-in spec path, durable state on disk, and gates that are code rather than prose.

1 8d ago A tokens not measured original MIT

mstack CLAUDE.md

03

romerma/mstack

Instructions file

Claude Code instructions for romerma/mstack, covering mstack, workflow and development.

1 8d ago A 284 tokens original MIT

implementer

04

romerma/mstack

Agent

Implements exactly one mstack work item, writes the tests that prove it, and records the requirement-to-test map. Never marks its own work approved or done.

1 8d ago A 36 tokens original MIT

orchestrator

05

romerma/mstack

Agent

Owns one work item from intake to close. Selects the next role, enforces the lifecycle, and monitors every launched agent to a terminal result. Never writes application code and never approves its own work.

1 8d ago A 46 tokens original MIT

reviewer

06

romerma/mstack

Agent

Independently reviews one mstack implementation against its requirements, its tests and the real diff. Runs the verification itself. Returns APPROVED or CHANGESREQUESTED and never edits the code.

1 8d ago A 41 tokens original MIT

spec-author

07

romerma/mstack

Agent

Writes the spec for one item under the mstack spec path. Produces proposal, design, tasks and requirements with stable R-ids, then hands off. Never reviews its own spec.

1 8d ago A 41 tokens original MIT

spec-reviewer

08

romerma/mstack

Agent

Reviews one mstack spec independently and adversarially before any code is written. Returns APPROVED or CHANGESREQUESTED. Never edits the spec and never reviews a spec it wrote.

1 8d ago A 41 tokens original MIT

SessionStart

09

romerma/mstack

Hook

Runs when a session starts, running an inline shell check. From romerma/mstack.

1 8d ago A tokens not measured original MIT

PostToolUse

10

romerma/mstack

Hook

Runs after a tool call finishes for Edit and Write tool calls, running an inline shell check. From romerma/mstack.

1 8d ago A tokens not measured original MIT

SubagentStop

11

romerma/mstack

Hook

Runs when a subagent finishes, running an inline shell check. From romerma/mstack.

1 8d ago A tokens not measured original MIT

Stop

12

romerma/mstack

Hook

Runs when the agent finishes a response, running an inline shell check. From romerma/mstack.

1 8d ago A tokens not measured original MIT

PreToolUse

13

romerma/mstack

Hook

Runs before the agent uses a tool for Bash tool calls, running an inline shell check. From romerma/mstack.

1 8d ago A tokens not measured original MIT

commit

14

romerma/mstack

Skill Claude CodeCodex

Write commit messages that hold to Conventional Commits 1.0.0, with the type the change actually earns and a description a stranger can read from the log alone. Use before every git commit, when amending or rewording one, when squashing, and when writing a PR title.

1 8d ago A 63 tokens original MIT

design

15

romerma/mstack

Skill Claude CodeCodex

Design the shape of a change before writing it, by generating structurally distinct candidates, judging them against criteria fixed in advance, and recording the decision with its rejected alternatives. Use before implementing anything non-trivial, or when asked to design, architect, or choose an approach.

1 8d ago A 57 tokens original MIT

implement

16

romerma/mstack

Skill Claude CodeCodex

Implement one work item with the tests that prove it, keeping durable state current and recording the requirement-to-test map, then hand it to a different pass to judge. Use when starting to write code for an item, or when asked to implement, build, or fix something under mstack.

1 8d ago A 60 tokens original MIT

orchestrate

17

romerma/mstack

Skill Claude CodeCodex

Run a program of work across many changes, tracks and sessions, with isolated worktrees, briefs a stranger could execute, and continuous landing. Use only for multi-session programs; single-session work should use the feature route instead.

1 8d ago A 49 tokens original MIT

reflect

18

romerma/mstack

Skill Claude CodeCodex

Review a finished session for what should change in the workflow itself, deciding for each lesson which document or check should absorb it. Use after closing an item, after something went wrong, or when asked to run a retro or improve the process.

1 8d ago A 50 tokens original MIT

review

19

romerma/mstack

Skill Claude CodeCodex

Judge work that already exists against its requirements, its tests and the real diff, using reviewers that did not write the code. Use to review a branch, a PR, or a diff, or when asked whether a change is ready.

1 8d ago A 49 tokens original MIT

mstack

20

romerma/mstack

Skill Claude CodeCodex

Start a task under the mstack workflow. Matches the request to a playbook, copies its steps into the todo list, and routes through understand, design, spec, implement, verify, review and ship. Use for /mstack, or when the user asks to work rigorously, to verify something properly, or to run a task end to end.

1 8d ago A 75 tokens original MIT

setup

21

romerma/mstack

Skill Claude CodeCodex

Initialize mstack in a repository, creating the durable state directory and seeding the work queue from what the repo already tells you. Use once per repository, when mstack reports no .mstack directory, or when asked to set up or install mstack here.

1 8d ago A 55 tokens original MIT

ship

22

romerma/mstack

Skill Claude CodeCodex

Take a verified change through review threads, CI and the merge gate to a merged state. Use to open or land a PR, to get a change merged, or when asked to ship or babysit something to green.

1 8d ago A 46 tokens original MIT

spec

23

romerma/mstack

Skill Claude CodeCodex

Write and adversarially review a specification before any code is written, with stable requirement ids, EARS statements and a task list that maps back to them. Use on the opt-in spec path, when an item is marked sdd, when a change crosses several subsystems, or when asked to write a spec or requirements.

1 8d ago A 68 tokens original MIT

understand

24

romerma/mstack

Skill Claude CodeCodex

Build a traced mental model of a subsystem before changing it, covering both how it works now and why it ended up this way. Use before designing or fixing anything in unfamiliar code, or when asked how something works, why a decision was made, or to explain a part of the codebase.

1 8d ago A 62 tokens original MIT