3 Commits

Include renames

• fix(agents): survive long overnight runs

  - compact at memory-safe token budget (~20k) not the 131k window; gpt-oss
    hits oMLX's prefill memory guard ~47k tokens on a 24GB machine
  - repair invalid_tool_calls (malformed JSON args) before echoing history —
    LangChain serializes the raw malformed string which oMLX 500s on
  - retry invokes (4x) with an oMLX watchdog restart between attempts
  - ALWAYS finalize: salvage picks from findings even after invoke failures
  
  Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
  Claude-Session: https://claude.ai/code/session_016DvdRZkyCy9NJpHYDdPqmS

  Aryan SinghandClaude Opus 4.8 2026-07-02 12:22:29 -05:00
  6b1c461a80
• Ready for first real test
  Aryan Singh 2026-07-02 00:46:34 -05:00
  b91c81ebdf
• feat(agents): research loop with budget guards, compaction, depth nudge, finalize

  Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
  Claude-Session: https://claude.ai/code/session_016DvdRZkyCy9NJpHYDdPqmS

  Aryan SinghandClaude Opus 4.8 2026-07-01 16:29:29 -05:00
  b2ae3d6ced