Requirements
Detailed, actionable requirements derived from your task description and grounded in your project's existing patterns and constraints.
AI Workloads
Turn ambiguous task tickets into precise, reviewable specs — and address every reviewer comment in a single pass. Kabori handles the definition work so your team can focus on the build.
01 · Task Refinement
Task Refinement uses AI to enrich raw task titles with structured requirements, acceptance criteria, edge cases, and implementation hints — automatically, before your team writes a single line of code.
## Requirements
- Enforce 100 req/min per authenticated user (sliding window)
- Return 429 with Retry-After header on breach
- Exempt service-to-service tokens from rate limiting
- Log rate-limit events to the audit trail within 1s
## Acceptance Criteria
- [ ] 101st request within 60s returns 429 with Retry-After
- [ ] Counter resets on window boundary (test with mock clock)
- [ ] Service tokens bypass limit and proceed normally
- [ ] Rate-limit events appear in audit log within 1s
## Edge Cases
- Concurrent requests exactly at the window boundary
- Requests with malformed or missing auth tokens
- Redis unavailable — fail open vs. fail closed decision needed
## Implementation Hints
- See src/middleware/throttle.ts for existing patterns
- Use shared Redis client at src/lib/redis.ts
Detailed, actionable requirements derived from your task description and grounded in your project's existing patterns and constraints.
A testable checklist that defines done — unambiguously. No more half-shipped features because the definition of complete was unclear.
Boundary conditions and failure modes surfaced before any code is written — saving hours of debugging and incident response later.
Unit, integration, and end-to-end testing scenarios tailored to the scope of your task — so QA and developers align before implementation begins.
Concrete pointers to relevant modules, patterns, and APIs in your codebase — so developers know exactly where to start.
Use custom templates and per-project instructions to control the spec format. Override settings per task, or mark specific tasks as exempt from automated refinement.
02 · Change Requests
Kabori reads your reviewer's comments, understands their intent, and applies all requested changes at once — turning review cycles from days into minutes.
Reviewer Comments
“Extract this into a helper function” — auth/session.ts:42
“Add error handling for the null case” — api/users.ts:87
“This should use the shared logger, not console.log” — services/payment.ts:15
“Missing test for the retry logic path” — services/payment.test.ts
Applied Changes
+ export function extractTokenValidator(token: string): ValidationResult
+ if (!user) throw new NotFoundError(`User $${id} not found`)
+ logger.info('Payment processed', { orderId, amount })
+ it('retries on transient failure', async () => { ... })
Kabori reads every reviewer comment in context — understanding which line it refers to, what the reviewer wants changed, and why.
Addresses all reviewer comments simultaneously — not one at a time. The resulting diff is coherent and ready for a single re-review.
Changes are minimal and targeted — only the lines that need to change are touched, keeping the diff clean and the review history clear.
Alongside the code changes, Kabori drafts replies to each reviewer comment — explaining what was changed and why, saving time on the back-and-forth.
Join the private beta and let Kabori handle the planning and refinement work — so your team spends time building, not clarifying.