| name | research |
| description | Deep research workflow for technologies, concepts, or complex topics |
Phase 0 — BOOTSTRAP
- Read
evolution.md in this skill's folder (if it exists). Apply accumulated lessons.
- Read
.agents/skills/orchestrate/SKILL.md and its references/inline-mode.md. You will follow orchestrate's full protocol with the supplementations below.
Strictly comply with the instructions in the orchestrate skill. The research skill serves merely as a decorator in Python; it is not the primary function.
Supplementation for Phase 1
- Explore agent - Should always include at least the following directories:
30_Research/, 40_Wiki/, 20_Project/.
Supplementation for Phase 2 (No Supplementation for Phase 1.5)
Integrate these parts into decomposition output:
- Web research — list specific search domains
- Main note →
30_Research/<Subfolder>/<Topic>.md with topic-appropriate sections (not a fixed template), always ending with ## References
- Wiki concepts → Spawn an agent with telling it to activate
/atomic-note to scan the main note.
- Canvas → yes/no (opt-in, delegates to
/json-canvas)
Supplementation for Phase 3
Constraints for implementers:
- References (MANDATORY, last section): local
[[Note]], online [Title](URL), every sourced claim cited, group if many
Supplementation for Phase 4
- Missing, empty or incomplete
## References section → hard-fail, needs-revision
Supplementation for Phase 5
Append to synthesis output:
Edge Cases
- Already covered: update/expand, not duplicate
- Follow-up: modify in place, no duplicates, update References