@maintainabilityai/research-runner 0.1.18 → 0.1.19

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -154,15 +154,13 @@ function formatForHuman(opts) {
154
154
  lines.push('Assign this issue to `@github-copilot` (or another agent) to produce the synthesis. The assignee should:');
155
155
  lines.push('');
156
156
  lines.push('1. **Read every source** above. The `S[N]` ids are how you cite them.');
157
- lines.push('2. **Open a PR** with a new file under `research/` containing the synthesis markdown.');
158
- lines.push('3. **Follow the canonical structure** see `.caterpillar/prompts/research/synthesis.md` in this mesh. The synthesis must have:');
159
- lines.push(' - 10 H2 sections in the order: `Executive Summary`, `Cross-Source Analysis`, `Jobs-to-be-Done Analysis`, `Whitespace Analysis`, `Patent Landscape`, `Community Signal`, `Academic Foundation`, `Conclusions`, `Recommendations`, `Open Questions`.');
160
- lines.push(' - Every claim cites at least one `S[N]`.');
161
- lines.push(' - Every Conclusion `C[N]` cites ≥2 sources (≥1 if confidence is LOW).');
162
- lines.push(' - Every Recommendation references at least one `C[N]`.');
157
+ lines.push('2. **Read the synthesis spec FIRST**: `.caterpillar/prompts/research/synthesis.md` in this mesh is the canonical structure. **Do NOT use any section names, ordering, or formatting other than what that file specifies.** The list of sections, the per-finding scaffolding, the citation rules, and the heading drift constraints all live there — follow them exactly.');
158
+ lines.push('3. **Open a PR** with a new file under `research/<run-id>/synthesis.md` containing the synthesis markdown.');
163
159
  lines.push('4. **PR labels** to apply: `research-synthesis`, `ai-assisted`.');
164
160
  lines.push('5. **Once merged**, the PRD agent will read your synthesis + the mesh + impacted code repos to produce per-repo landing issues.');
165
161
  lines.push('');
162
+ lines.push('> ⚠ Earlier iterations of this comment listed sections inline, which let agents skip the spec file. The spec at `.caterpillar/prompts/research/synthesis.md` is now the **single source of truth** for sections, ordering, citation rules, and per-finding format. The downstream PRD agent parses on exact heading strings — drift breaks the pipeline.');
163
+ lines.push('');
166
164
  lines.push('---');
167
165
  lines.push('');
168
166
  lines.push(`🤖 Generated by \`research-runner archeologist\`. Run id: \`${runId}\` (see the Hatter's Tag for agent version + audit chain).`);
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@maintainabilityai/research-runner",
3
- "version": "0.1.18",
3
+ "version": "0.1.19",
4
4
  "description": "Research + PRD agent runner — orchestrates the Archeologist and PRD pipelines for the MaintainabilityAI governance mesh",
5
5
  "license": "MIT",
6
6
  "author": "MaintainabilityAI",