What's Changed
Deep Executor: Hephaestus-Level Capabilities (#264)
Comprehensive upgrade to the deep-executor agent based on research comparing oh-my-opencode's Hephaestus agent:
| # | Feature | Priority |
|---|---|---|
| 1 | Parallel exploration via explore/researcher delegation | P1 |
| 2 | Structured escalation after 3 failures → architect-medium | P1 |
| 3 | Mandatory code style pattern discovery before coding | P1 |
| 4 | Communication efficiency rules (forbidden phrases) | P2 |
| 5 | Adaptive reasoning by task complexity | P2 |
| 6 | Expanded 8-item verification checklist | P3 |
Files Changed
agents/deep-executor.md— Prompt expanded from ~194 to ~365 linessrc/agents/deep-executor.ts— AddedTasktool for explore/researcher delegationANALYSIS.md— Full 20-section Hephaestus vs Deep-Executor comparisonresearch/hephaestus-vs-deep-executor-comparison.md— Feature gap matrix
Full Changelog: v3.9.2...v3.9.3