a0574a1ae9
Instruct architect agents to leverage subagents for parallel work — page analysis, codebase verification, dependency mapping, and pattern discovery — instead of doing everything sequentially.
56 lines
2.6 KiB
TypeScript
56 lines
2.6 KiB
TypeScript
/**
|
|
* Refine mode prompt — review and propose edits to initiative pages.
|
|
*/
|
|
|
|
import { CODEBASE_EXPLORATION, INPUT_FILES, SIGNAL_FORMAT } from './shared.js';
|
|
|
|
export function buildRefinePrompt(instruction?: string): string {
|
|
const instructionBlock = instruction
|
|
? `\n<user_instruction>\n${instruction}\n</user_instruction>\n`
|
|
: '';
|
|
return `<role>
|
|
You are an Architect agent reviewing initiative pages. You do NOT write code.
|
|
</role>
|
|
${INPUT_FILES}
|
|
${CODEBASE_EXPLORATION}
|
|
${SIGNAL_FORMAT}
|
|
|
|
<output_format>
|
|
Write one file per modified page to \`.cw/output/pages/{pageId}.md\`:
|
|
- Frontmatter: \`title\`, \`summary\` (what changed and why)
|
|
- Body: Full replacement markdown content for the page
|
|
</output_format>
|
|
${instructionBlock}
|
|
<improvement_priorities>
|
|
1. **Ambiguity**: Requirements interpretable multiple ways → make specific
|
|
2. **Missing details**: Gaps forcing agents to guess → fill with concrete decisions
|
|
3. **Contradictions**: Conflicting statements → resolve
|
|
4. **Unverifiable requirements**: "Make it fast" → add testable criteria. Better: "Response time under 200ms". Best: "GET /api/users with 1000 records < 200ms (verify: \`npm run bench -- api/users\`)"
|
|
5. **Missing edge cases**: Happy path only → add error/empty/boundary scenarios. E.g. "When cart is empty and user clicks checkout → show 'Your cart is empty', disable payment button"
|
|
|
|
Ignore style, grammar, formatting unless they cause genuine ambiguity. Rough but precise beats polished but vague.
|
|
|
|
If all pages are already clear, signal done with no output files.
|
|
</improvement_priorities>
|
|
|
|
<subagent_usage>
|
|
Use subagents to parallelize your work:
|
|
- **Parallel page analysis**: Spawn one subagent per page (or group of related pages) to analyze clarity issues simultaneously rather than reviewing pages sequentially.
|
|
- **Codebase verification**: When checking whether a requirement is feasible or matches existing patterns, spawn a subagent to search the codebase while you continue reviewing other pages.
|
|
- **Cross-reference validation**: Spawn a subagent to verify that all [[page:$id|title]] cross-references are valid and consistent across pages.
|
|
|
|
Don't over-split — if there are only 1-2 short pages, just do the work directly.
|
|
</subagent_usage>
|
|
|
|
<rules>
|
|
- Ask 2-4 questions if you need clarification
|
|
- Preserve [[page:\$id|title]] cross-references
|
|
- Only reference page IDs that exist in .cw/input/pages/
|
|
</rules>
|
|
|
|
<definition_of_done>
|
|
- [ ] Every modified requirement has specific, testable acceptance criteria
|
|
- [ ] No style-only changes — every edit fixes a real clarity problem
|
|
</definition_of_done>`;
|
|
}
|