Skip to content
Skill Authoring
Skill

/governor

Use when the user says 'new project', 'project init', 'what tier', 'scope', or discusses project maturity, complexity budget, or what's appropriate to build.

BOOST
From plugin
memstack
42587 skills2 commands4 hooks
Install
$ npx -y skills add cwinvestments/memstack --skill governor --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/governor

Context preview

The summary Claude sees to decide when to auto-load this skill.

Use when the user says 'new project', 'project init', 'what tier', 'scope', or discusses project maturity, complexity budget, or what's appropriate to build.

SKILL.md

governor.SKILL.md
name: governor
description: "Use when the user says 'new project', 'project init', 'what tier', 'scope', or discusses project maturity, complexity budget, or what's appropriate to build."
version: 1.0.0

🏛️ Governor: Portfolio Governance

*Enforce tier-appropriate complexity. Prevent over-engineering the #1 waste of time in AI-assisted development.*

Activation

When this skill activates, output:

`🏛️ Governor: Checking project tier constraints...`

Then execute the protocol below.

Context Guard

| Context | Status | Priority | |---------|--------|----------| | **User starts a new project ("new project", "init", "scaffold")** | ACTIVE: assign tier | P1 | | **User asks "what tier", "what's allowed", "scope check"** | ACTIVE, report current tier constraints | P1 | | **User proposes work that exceeds current tier** | ACTIVE: flag and advise | P2 | | **User is executing work within tier constraints** | DORMANT: don't interrupt | none | | **User explicitly overrides ("I know, do it anyway")** | DORMANT: user has authority | none |

Anti-Rationalization

If you're thinking any of these, STOP, you're about to let scope creep happen:

| You're thinking... | Reality | |---|---| | "Adding tests is always good practice" | Not for prototypes. Tests for throwaway code waste time. | | "This needs proper auth" | Single-user tools don't need auth. Add it when there are users. | | "Let me add CI/CD while I'm at it" | CI/CD for a prototype is gold-plating. Ship first. | | "Error handling should be comprehensive" | Prototype error handling = crash and log. That's it. | | "I should add monitoring" | < 10 users? Console.log is your monitoring. | | "This should be configurable" | Hardcode it. Make it configurable when someone asks. |

Protocol

Step 1: Determine Project Tier

Ask or infer the project tier from context:

| Tier | Description | Effort Allocation | |------|-------------|-------------------| | **Prototype** | Exploring an idea. May be thrown away. | Minimal: working code only | | **MVP** | Validated idea, building for first users. | Moderate: basic quality gates | | **Production** | Serving real users, needs reliability. | Full: complete quality stack |

If tier is unclear, default to **Prototype** and escalate only when evidence suggests otherwise.

Step 2: Apply Tier Constraints

Prototype: Move Fast, Break Things

| Allowed | NOT Allowed | |---------|-------------| | Working code that demonstrates the idea | Unit tests | | Hardcoded config values | CI/CD pipelines | | Console.log for debugging | Type systems / strict typing | | Single-file scripts | Monitoring / alerting | | README with setup instructions | Authentication / authorization | | | Infrastructure-as-code | | | Rate limiting | | | Database migrations (use SQLite) |

**Prototype rule:** If it works in a demo, ship it.

MVP: Prove It Works

| Allowed | NOT Allowed | |---------|-------------| | Everything from Prototype, plus: | Integration test suites | | Basic unit tests (happy path only) | Full CI/CD with staging | | Simple error handling (try/catch at boundaries) | Monitoring dashboards | | Environment variables for config | Multi-environment deploys | | Basic input validation | Performance optimization | | Simple auth (if multi-user) | Horizontal scaling | | README + basic API docs | Comprehensive logging |

**MVP rule:** If the first 10 users can use it reliably, ship it.

Production: Reliability Matters

| Allowed | Required | |---------|----------| | Everything from MVP, plus: | Comprehensive tests (unit + integration) | | Performance optimization | CI/CD pipeline | | Monitoring and alerting | Error tracking (Sentry or equivalent) | | Multi-environment deployment | Input validation at all boundaries | | Horizontal scaling | Authentication + authorization | | Database migrations | Logging with structured output | | Rate limiting | API documentation |

**Production rule:** If it breaks at 3 AM, someone gets paged.

Step 3: Report Constraints

Output a brief summary:

🏛️ Project: {name}
   Tier: {Prototype | MVP | Production}
   Allowed: {brief list}
   NOT allowed: {brief list of key restrictions}

Step 4: Flag Violations

When the user proposes work that exceeds the tier, flag it:

🏛️ Governor, Scope check:
   You're proposing {X}, but this is a {Tier} project.
   {X} is a {higher tier} concern. Current tier doesn't require it.
   Want to proceed anyway, or skip it for now?

Always defer to the user if they override. Governor advises, doesn't block.

Anti-Patterns by Tier

Prototype Anti-Patterns: DON'T DO THIS

1. **Writing tests for throwaway code**: If the prototype proves the idea wrong, those tests are wasted 2. **Adding auth to single-user tools**: You're the only user. Skip it 3. **Setting up CI/CD**: You're not deploying to production. `git push` is your CI 4. **Using TypeScript for a quick script**: JavaScript is fine for prototypes 5. **Adding rate limiting**: You have 0 users. Rate limit when you have 10 6. **Creating database migrations**: SQLite + direct schema changes. Migrate when you scale 7. **Building admin dashboards**: Database GUI tool (TablePlus, DBeaver) is your admin panel 8. **Over-abstracting**: 3 similar lines > 1 premature abstraction 9. **Adding comprehensive error handling**: Crash and read the stack trace. That's debugging 10. **Monitoring and alerting**: Console output is your monitoring

MVP Anti-Patterns: DON'T DO THIS

1. **Integration test suites**: Happy-path unit tests are enough at MVP 2. **Multi-environment deploys**: One environment. Dev IS production 3. **Performance optimization**: Make it work, make it right, THEN make it fast. You're at step 2 4. **Horizontal scaling**: Vertical scale (bigger server) until proven insufficient 5. **Comprehensive logging**: Log errors and key events. Not every function call

Production Anti-Patterns: DON'T DO THIS

1. **Skipping tests to "move faster"**: You'll mov

Read more
Ships withmemstack

The structured skill framework for Claude Code: 131 professional skills for deployment, security, databases, content, marketing, and more. Skills activate automatically when you need them. Say "deploy this to Railway" and the right skill loads on demand.

Get the whole plugin

Other skills on memstack.