Use Case: Single Skill

When to use this path

Use this when the output is narrowly scoped and mostly owned by one domain:

• one UI component
• one API endpoint
• one bug in one layer
• one refactor in one module

If the task requires cross-domain coordination (API contract + UI + QA), use Multi-Agent Project.

Preflight checklist

Before prompting, define:

1. exact output (file or behavior)
2. target stack and versions
3. acceptance criteria
4. test expectations

Prompt template

Build <specific artifact> using <stack>.
Constraints: <style/perf/security constraints>.
Acceptance criteria:
1) ...
2) ...
Add tests for: <critical cases>.

Example prompt

Create a login form component in React + TypeScript + Tailwind CSS.
Constraints: accessible labels, client-side validation, no external form library.
Acceptance criteria:
1) email and password validation messages
2) disabled submit while invalid
3) keyboard and screen-reader friendly
Add unit tests for valid/invalid submit paths.

Expected execution flow

1. The relevant skill is auto-selected.
2. The agent proposes implementation and assumptions.
3. You confirm or adjust assumptions.
4. The agent ships code and tests.
5. You run local verification and request small follow-ups.

Quality gate before merge

• behavior matches acceptance criteria
• tests cover happy path and core edge cases
• no unrelated file changes
• no hidden breaking changes to shared modules

Escalation signals

Switch to multi-agent flow when:

• UI work requires new API contracts
• one fix creates cascading changes across layers
• scope grows beyond one domain after first iteration

Done criteria

Single-skill execution is done when:

• target artifact is implemented
• acceptance criteria are demonstrably satisfied
• tests are added or updated for the changed behavior