Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 12 additions & 0 deletions .changeset/sync-agents-md-architecture.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,12 @@
---
"create-expert": patch
---

Sync AGENTS.md template with actual PBT framework architecture

The architecture diagram now correctly reflects the implementation:
- Replaced non-existent experts (happy-path-manager, unhappy-path-manager,
adversarial-manager, expert-designer) with actual implementation
- Added ecosystem-builder, integration-manager, functional-manager,
usability-manager
- Removed references to non-existent `think` tool from examples
25 changes: 11 additions & 14 deletions apps/create-expert/src/lib/agents-md-template.ts
Original file line number Diff line number Diff line change
Expand Up @@ -36,16 +36,13 @@ This project uses a Property-Based Testing (PBT) approach to create high-quality

\`\`\`
create-expert (Coordinator)
├── property-extractor → Extracts success criteria
├── happy-path-manager → Stage 1: Normal operation testing
│ ├── expert-designer → Creates/improves Expert definitions
│ └── expert-tester → Runs tests and evaluates results
├── unhappy-path-manager → Stage 2: Error handling testing
│ ├── expert-designer
│ └── expert-tester
├── adversarial-manager → Stage 3: Security testing
│ ├── expert-designer
│ └── expert-tester
├── property-extractor → Extracts testable properties from requirements
├── ecosystem-builder → Creates Expert ecosystem (main + demo + setup + doctor)
├── integration-manager → Orchestrates functional and usability testing
│ ├── functional-manager → Runs happy-path, unhappy-path, adversarial tests
│ │ └── expert-tester → Executes tests and reports results
│ └── usability-manager → Runs demo, setup, doctor, error guidance tests
│ └── expert-tester → Executes tests and reports results
└── report-generator → Final property achievement report
\`\`\`

Expand Down Expand Up @@ -123,7 +120,7 @@ Detailed instructions for the expert.
type = "mcpStdioSkill"
command = "npx"
packageName = "@perstack/base"
pick = ["readTextFile", "think", "attemptCompletion"]
pick = ["readTextFile", "attemptCompletion"]
\`\`\`

## Best Practices for Creating Experts
Expand Down Expand Up @@ -175,7 +172,7 @@ Always use \`pick\` to limit tools:
type = "mcpStdioSkill"
command = "npx"
packageName = "@perstack/base"
pick = ["readTextFile", "think", "attemptCompletion"] # Only what's needed
pick = ["readTextFile", "attemptCompletion"] # Only what's needed
\`\`\`

### 4. Let Them Collaborate
Expand Down Expand Up @@ -276,14 +273,14 @@ endpoint = "https://api.example.com/mcp"
\`@perstack/base\` provides essential tools:
- File operations: \`readTextFile\`, \`writeTextFile\`, \`editTextFile\`, etc.
- Shell execution: \`exec\`
- Control flow: \`attemptCompletion\`, \`think\`, \`todo\`
- Control flow: \`attemptCompletion\`, \`todo\`

\`\`\`toml
[experts."my-expert".skills."@perstack/base"]
type = "mcpStdioSkill"
command = "npx"
packageName = "@perstack/base"
pick = ["readTextFile", "writeTextFile", "think", "attemptCompletion"]
pick = ["readTextFile", "writeTextFile", "attemptCompletion"]
\`\`\`

## Testing Experts
Expand Down