DeepSeekHarness update adds multimodal support and Codex sub-agents
智东西 · wechat · 2026-08-20
DeepSeekHarness version v0.1.0-rc.8 introduces significant multimodal capabilities, supporting native image requests and mixed text-image inputs for commands like /goal and /plan. The update integrates ClaudeCode and Codex into the sub-agent ecosystem, supporting non-interactive modes and multiple named instances for Codex. For image processing, the system allows visual models to input images directly, while text-only models can use tools like OCR, color statistics, and pixel scanning to interpret images indirectly. It also fixes Windows terminal issues, image request failures, and streaming generation bugs, alongside SQLite backend performance improvements.
More from coding & agent
- Six-month prediction: hiring for generic agent harnesses built from scratch will cool — yi111 · 2026-10-16
- Omarchy ISO wouldn't boot on NVIDIA, so he had Claude patch it — indragie · 2026-10-16
- Dev uses Grok to clean up 3D printer firmware, catches it sneaking in nonsense code — yacineMTB · 2026-10-16
- A rude AI agent took 89% of a shared resource in 102-run multi-agent personality experiment — Brave-History-6502 · 2026-10-16
- Guardrails can make risky delegation work — but it takes a full-time team, says iandanforth — iandanforth · 2026-10-16
- What can you truly delegate? Only tasks with unambiguous, verifiable results — iandanforth · 2026-10-16