Quickstart#

This quickstart uses the current CafeKit runtime:

/hapo:specs -> /hapo:develop -> /hapo:test -> /hapo:code-review

1. Install CafeKit#

npx @haposoft/cafekit

2. Create A Spec#

Describe the feature in plain language:

/hapo:specs Build a meeting transcript extension with AI summaries

CafeKit will generate a spec workspace like:

specs/meet-transcript-mvp/
├── spec.json
├── requirements.md
├── research.md
├── design.md
└── tasks/
    ├── task-R0-01-*.md
    ├── task-R0-02-*.md
    └── ...

3. Validate The Spec#

Before coding, run the spec validation pass:

/hapo:specs --validate meet-transcript-mvp

This step is where CafeKit checks:

• requirements, design, and tasks agree with each other
• contracts are explicit
• the spec is ready for implementation

4. Implement One Task#

The safest pattern is task-by-task execution:

/hapo:develop meet-transcript-mvp task-R0-02-auth-setup-dual-mode.md

CafeKit will:

1. load the task packet
2. enforce task boundaries
3. run the quality gate
4. sync task state only after verification passes
5. run a task-level docs checkpoint

5. Run Verification#

/hapo:test --full

CafeKit reports a structured verdict such as:

• PASS
• FAIL
• PARTIAL
• NO_TESTS

Important:

• NO_TESTS is not equivalent to success
• preflight failures should block completion

6. Run Review#

/hapo:code-review --pending

Use this to verify correctness, regressions, and security after implementation.

7. Audit Task State When Needed#

/hapo:sync audit meet-transcript-mvp

Use sync audit when:

• task files and spec.json drift
• a run was interrupted
• you want to confirm the next unblocked task

Recommended Working Pattern#

For most features:

spec -> validate -> develop one task -> test -> review -> next task

This keeps task state, evidence, and docs aligned.