Issues

• Upgrade the Anthropic SDK, OpenAI SDK and commander majors deliberately

  housekeepingP2
  agentmuxai/muxcode#47 · Filter by author agent3-workflow opened on Sep 26, 2026

• Tracking: make Mux Code a working AgentMux harness, then close the gaps with current CLI agents

  tracking
  agentmuxai/muxcode#33 · Filter by author agent3-workflow opened on Sep 26, 2026

• Publish v0.2.0 to npm (needs the npm org and NPM_TOKEN)

  housekeepingP0
  agentmuxai/muxcode#32 · Filter by author agent3-workflow opened on Sep 26, 2026

• Reconcile main with the default branch, and fix package.json entry points

  housekeepingP1
  agentmuxai/muxcode#31 · Filter by author agent3-workflow opened on Sep 26, 2026
  1

• Secret redaction and prompt-injection defences

  enhancementP3phase-3
  agentmuxai/muxcode#30 · Filter by author agent3-workflow opened on Sep 26, 2026

• Sub-agents

  enhancementP3phase-3
  agentmuxai/muxcode#29 · Filter by author agent3-workflow opened on Sep 26, 2026

• Checkpoints and rewind

  enhancementP3phase-3
  agentmuxai/muxcode#28 · Filter by author agent3-workflow opened on Sep 26, 2026

• Skills, slash commands and AskUserQuestion

  enhancementP2phase-3
  agentmuxai/muxcode#27 · Filter by author agent3-workflow opened on Sep 26, 2026

• Persistent mode: --input-format stream-json

  enhancementP3phase-3
  agentmuxai/muxcode#26 · Filter by author agent3-workflow opened on Sep 26, 2026

• Permission modes, Claude-syntax rules, and a destructive-command guard

  enhancementP2phase-3
  agentmuxai/muxcode#25 · Filter by author agent3-workflow opened on Sep 26, 2026

• Run Claude-compatible hooks (PreToolUse Bash, PreCompact, …)

  enhancementP2phase-3
  agentmuxai/muxcode#24 · Filter by author agent3-workflow opened on Sep 26, 2026

• Speculative decoding and model routing

  enhancementP3phase-2
  agentmuxai/muxcode#23 · Filter by author agent3-workflow opened on Sep 26, 2026

• Fix llama-server release asset names and first-run download

  enhancementP2phase-2
  agentmuxai/muxcode#22 · Filter by author agent3-workflow opened on Sep 26, 2026

• Refresh the model catalog, with checksums and context sizes

  enhancementP2phase-2
  agentmuxai/muxcode#21 · Filter by author agent3-workflow opened on Sep 26, 2026

• A small tool set for small contexts (tool search)

  enhancementP1phase-2
  agentmuxai/muxcode#20 · Filter by author agent3-workflow opened on Sep 26, 2026

• Robust tool calling on open-weight models

  enhancementP1phase-2
  agentmuxai/muxcode#19 · Filter by author agent3-workflow opened on Sep 26, 2026

• Context size from GGUF metadata, and current llama-server flags

  enhancementP1phase-2
  agentmuxai/muxcode#18 · Filter by author agent3-workflow opened on Sep 26, 2026

• Reuse the KV cache across turns

  enhancementP1phase-2
  agentmuxai/muxcode#17 · Filter by author agent3-workflow opened on Sep 26, 2026

• Keep llama-server alive across turns

  enhancementP1phase-2
  agentmuxai/muxcode#16 · Filter by author agent3-workflow opened on Sep 26, 2026

• MCP hygiene: namespacing, isError, images, http transport, ${VAR}

  enhancementP2phase-1
  agentmuxai/muxcode#15 · Filter by author agent3-workflow opened on Sep 26, 2026

• Honest stop reasons, retries with backoff, loop detection

  enhancementP1phase-1
  agentmuxai/muxcode#14 · Filter by author agent3-workflow opened on Sep 26, 2026

• Context accounting and auto-compaction, with compact_boundary

  enhancementP1phase-1
  agentmuxai/muxcode#13 · Filter by author agent3-workflow opened on Sep 26, 2026

• A real system prompt and prompt caching

  enhancementP1phase-1
  agentmuxai/muxcode#12 · Filter by author agent3-workflow opened on Sep 26, 2026

• Cap tool output and spill the rest to a file the model can read

  enhancementP1phase-1
  agentmuxai/muxcode#11 · Filter by author agent3-workflow opened on Sep 26, 2026

• Tolerant edit matching, apply_patch for OpenAI models, read-before-edit guard

  enhancementP1phase-1
  agentmuxai/muxcode#10 · Filter by author agent3-workflow opened on Sep 26, 2026