I created a claude code deep researcher that seems to work better than the current deep research models
This repository also configures its own agents. See what Claude-Code-Deep-Research tells them →
AnkitClassicVision/Claude-Code-Deep-Research
Agent Claude Code
Use this agent to assemble the final report files in Phase 5 and to apply post-gate fixes in Phase 6 of deep research. The editor formats and structures; it never adds facts.
AnkitClassicVision/Claude-Code-Deep-Research
Agent Claude Code
Use this agent to convert scout branch notes into evidence ledger rows. Run at every checkpoint during Phase 3 of deep research, before the convergence log is updated.
AnkitClassicVision/Claude-Code-Deep-Research
Agent Claude Code
Use this agent on the draft report during Phase 5 of deep research, after synthesis and before the gates. Mandatory for Deep and Exhaustive tiers and for any decision-tier claim.
AnkitClassicVision/Claude-Code-Deep-Research
Agent Claude Code
Use this agent when sources conflict on a finding-tier or decision-tier claim, or when stoprule.py returns ESCALATESTUCK. Classifies the conflict and picks a canonical value or declares it unresolved.
AnkitClassicVision/Claude-Code-Deep-Research
Agent Claude Code
Use this agent to execute one research branch: search, fetch, and summarize sources for a single hypothesis. Launch one scout per branch, in parallel. Use proactively during Phase 3 of deep research.
AnkitClassicVision/Claude-Code-Deep-Research
Agent Claude Code
Use this agent to verify finding-tier and decision-tier claims in the evidence ledger during Phase 4 of deep research. Must run on a different model than the one writing the synthesis.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: