mstack
Skill0 starsChanged 27 days agoMITView on GitHub
Michael's engineering workflow bundle. Explicit entry point for planning, implementation, investigation, review, and PR delivery.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- VersoXBT/mstacktheir own version, first seen May 5, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- VersoXBT/mstack★ 4same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository8 other files it gives its agents.
- AGENTS.mdAGENTS.md
- agent-orchestratorSkill
- classify-code-reviewSkill
- conventional-commitsSkill
- go-test-effectivenessSkill
- service-designSkill
- show-complexitySkill
- spec-plannerSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 16
If an explicit @mstack invocation cannot load this entry, report that rather than substituting another skill.
- Distinctive: line 82
Read relevant principles before applying them within the task's scope.
- Very specific: line 103
Use question for consequential decisions, with a recommendation and meaningful alternatives.
- Distinctive: line 107
Use the available Plannotator tools when the user requests artifact review.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Mstack
Select the matching playbook and briefly name it. Playbooks define procedures; principles guide decisions. Load supporting files only as needed, resolving relative links from the containing file.
Track the playbook’s completion conditions in a checklist. If no playbook fits, define a task-specific checklist with an explicit deliverable and stopping condition.
For delegation or independent review, read orchestration.
If an explicit @mstack invocation cannot load this entry, report that rather than substituting another skill.
Playbooks
| Request | Playbook |
| New or changed behavior | Feature |
| Repair a reported defect | Bug fix |
| Behavior-preserving restructuring | Refactoring |
| New proposal, sketch or spec | Planning |
| Refine an existing proposal | Spec refinement |
| Implement a spec | Spec implementation |
| Read-only investigation of code or architecture | Investigation |
| Prototype or visual-parity work | Prototype |
| Performance measurements, fixes, or hillclimbing | Performance |
| Diagnose a live symptom or captured trace | Forensics |
| Independent review or blast-radius analysis | Review |
| Architecture, competing designs, or partitioned exploration | Exploration |
| Publish, rewrite, triage, fix feedback, inspect checks, or monitor | PR workflows |
| Rebase a PR or tidy merged branches and worktrees | PR workflows |
| Explicit stack or merge | Advanced delivery |
| Long active-session task or project-scale run | Autonomous run |
| Keep or audit a reviewable decision trail | Decision trail |
| Resume prior work | Pickup |
| Hand off or pause work | Pause |
| Create a project verification skill | Create |
| Maintain a project verification skill | Maintain |
| Stress-test a proposal through questions | Grilling |
| Clarify domain terms or record a glossary/decision | Domain modeling |
| Explain with a diagram or code-shape sketch | Show me |
| Explain in HTML with large visuals and minimal text | Big pictures, little words |
| Walk a reader through a change or data model as a shared tour | Tour |
| Assess complexity without changing code | Show complexity |
| Review recent work and live state | Recall |
| Retro the current session for skill lessons | Reflect |
| Summarize progress in the current conversation | Catch up |
| Clean up the scoped change (code, prose, comments) | Code cleanup |
| Write or review technical documentation | Technical writing |
| Improve prose | Prose cleanup |
| Review comment value | Comments |
| Find, apply, or survey project coding standards | Coding standards |
| Capture lessons about project code as rules or checks | Capture |
| Explain mstack usage or dependencies | Help |
Investigation
Establish the question and the decision it informs. Remain read-only unless the user separately requests an artifact or implementation.
- Runtime mechanics and ownership: use how.
- Design rationale and historical constraints: use why.
- Choosing between design options: use Architecture, not Why.
- Timing trends, slow tests, and benchmarks: use performance, read-only until implementation is approved.
- A teaching conversation: use teach.
- Essential versus accidental complexity: use show-complexity.
Trace the important path, inspect material failure cases, and cite the actual source. Delegate independent deep evidence gathering when useful. Stop once the question has a supported answer or a deciding fact is explicitly unavailable. Recommend a next action only when it follows from the findings; an investigation does not automatically become a prototype, refactor, or PR.
Shared rules
- Scope: Mstack applies to the invoked task. The selected playbook defines its work and stopping condition; reading another file supplies guidance, not authority to start another workflow. Explicit user instructions and repository rules govern the task. Preserve unrelated work and the selected delivery branch. A branch whose PR has merged or closed is no longer a delivery branch.
- Approval: Implementation, including code-writing experiments, requires an approved sketch or spec. Reuse an agreed sketch, spec, or feedback fix; seek renewed approval when new evidence materially changes the outcome. Read-only investigation, requested planning, and diff inspection need no implementation approval. Keep experiments in isolated, explicitly disposable targets. Before writing code, find and apply project standards.
- Delivery: Approved coding work starts in a prepared workspace. Approved coding playbooks authorize scoped commits, pushes, one PR, and required-CI repair. Ordinary delivery stops before merging. Stacks and merging require explicit requests.
- Completion: Support completion claims with observed evidence against the requested outcome; use verification for implementation and review. Report unmet requirements and unavailable or unrun checks. When blocked, preserve a resumable state and seek help rather than changing the goal or silently weakening a requirement.
Principles
Read relevant principles before applying them within the task's scope. Cite each principle that materially changed a decision by name and link, explaining the choice it changed.
| When deciding | Read |
| Core data, states, signatures, or ownership | Model the domain |
| Validation, errors, framework or network adapters | Boundary discipline |
| Refactoring, abstraction, deletion, or reader burden | Minimize reader load |
| Integrating a requirement or repeating failed fixes | Question the premise |
| User-visible behavior and product tradeoffs | Experience first |
| Novel designs with consequential uncertainty | Explore alternatives |
| Declaring a task complete | Prove it works |
| Designing or judging tests | Test behavior |
| Benchmarks, performance claims, or eval scores | Explain the number |
| Multi-step work, commits, migrations, or stacks | Verifiable units |
| Parallel writers or shared mutable state | Separate ownership |
| Large evidence, exploration, or handoffs | Guard context |
| Retryable commands and partial failure | Idempotent operations |
| Recurring mistakes or repeated manual work | Encode lessons |
Tools
Use question for consequential decisions, with a recommendation and meaningful alternatives; fall back to conversation when unavailable. Look up facts rather than asking the user to supply them.
Discover deferred tools through the available catalog rather than guessing tool names.
Use the available Plannotator tools when the user requests artifact review.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

