.agent/skills/protocols/. Unlike workflows (which execute tasks), protocols guide how you think about problems.
What is a Protocol?
A protocol is a documented thinking pattern that you can invoke repeatedly. It answers questions like:- “How do I evaluate this decision?”
- “What framework should I use for this analysis?”
- “What questions should I ask before acting?”
- Protocol 123 (Einstein Protocol): Spend 11x more time defining the problem than solving it
- Protocol 137 (Graph of Thoughts): Generate solution paths, score them, converge on the best
- Protocol 416 (Agent Swarm): Parallelize work across multiple AI agents
Protocol vs Workflow
| Aspect | Workflow | Protocol |
|---|---|---|
| Purpose | Execute tasks | Guide thinking |
| Location | .agent/workflows/ | .agent/skills/protocols/ |
| Invocation | Slash command (/deploy) | Referenced by ID (Protocol 123) |
| Format | Checklist of steps | Framework + decision logic |
| Example | ”Deploy to staging" | "55/5 Problem Definition Rule” |
Protocol Structure
1. YAML Frontmatter
2. Header Block
3. Core Principle
State the fundamental insight:4. The Trigger (When to Use)
Define clear activation conditions:5. The Mechanism (How It Works)
Provide step-by-step instructions:6. Example Application
7. Anti-Patterns
Step-by-Step: Writing Your First Protocol
Identify a recurring decision pattern
Look for situations where you repeatedly use the same thinking framework. Examples:
- Evaluating feature requests
- Deciding whether to refactor code
- Choosing between technical approaches
- Assessing project viability
Choose a protocol ID
Pick a number that fits the category:
- 0-99: Core principles
- 100-199: Decision frameworks
- 200-299: Verification methods
- 300-399: Workflow patterns
- 400-499: Architecture patterns
.agent/skills/protocols/decision/123-problem-definition.mdReal-World Protocol: Einstein Protocol (55/5 Rule)
Here’s how Protocol 123 is structured in the Athena codebase:Protocol Categories
Organize protocols by domain:Decision Frameworks
- Protocol 123: Einstein Protocol (55/5 Rule)
- Protocol 124: SDR Diagnosis
- Protocol 111: Premise Audit
Workflow Patterns
- Protocol 416: Agent Swarm (Parallel Execution)
- Protocol 419: Handoff Loop
- Protocol 137: Graph of Thoughts
Verification Methods
- Protocol 171: Cross-Model Validation
- Protocol 141: Claim Atomization Audit
- Protocol 105: Neuro-Symbolic Verification
Engineering
- Protocol 175: TDD Workflow
- Protocol 44: Micro-Commit Protocol
Best Practices
Start with 'why'
Every protocol should clearly state the problem it solves and the principle behind the solution.
Make it actionable
Include concrete steps, not just philosophy. Someone should be able to execute the protocol without asking for clarification.
Provide examples
Show the protocol in action with real scenarios. Abstract frameworks are hard to apply.
Version it
Use frontmatter to track creation and update dates. Protocols evolve as you learn.
Protocol Naming Conventions
Format:Protocol [ID]: [Name]
ID Ranges:
- 0-99: Core principles (Laws, Identity)
- 100-199: Decision & analysis
- 200-299: Verification & quality
- 300-399: Workflow & process
- 400-499: Architecture & systems
- 500+: Domain-specific
[id]-[slug].md
Examples:
123-einstein-protocol.md416-agent-swarm.md175-tdd-workflow.md
Integration with Workflows
Protocols are referenced within workflows:Common Pitfalls
| ❌ Don’t | ✅ Do Instead |
|---|---|
| Write philosophy without steps | Provide concrete, executable instructions |
| Skip the “when to use” section | Define clear activation triggers |
| Forget examples | Show the protocol solving a real problem |
| Use vague language | Be specific and measurable |
| Ignore anti-patterns | Document what NOT to do |
Next Steps
Semantic Search
Learn how Athena finds and retrieves your protocols using triple-path retrieval
Best Practices
Operational discipline for maintaining your protocol library