Commit History

  1. Fix: Argentine Spanish -> Mexican Spanish (escribís->escribes, tenés->tienes, Recordá->Recuerda, etc)

    Verified signature
    29d2d27
  2. v11: Fix thinking leak - ALWAYS cleanThinkingArtifacts, more aggressive detection, expanded patterns

    Verified signature
    2bac7f7

  1. v5.8.9: Fix dedup, upgrade Qwen3.6-35B-A3B, humanize v2, personality v2, /messages endpoint

    Z User
    b4c36fb
  2. v5.8.8c: add humanize to /chat endpoint, filter thinking artifacts

    Z User
    3438538
  3. v5.8.8b: fix /chat prompt with full identity + security rules

    Z User
    3f0356d
  4. v5.8.8: fix thinking leak, anti-manipulation compliance, hallucination of server data

    Z User
    cde5d07
  5. Revert "temp: add raw-config endpoint for migration"⁠

    Z User
    36af2b6
  6. temp: add raw-config endpoint for migration

    Z User
    f51940c
  7. fix: GLM API 1213 error - validate messages array before sending

    Z User
    539d7e9
  8. v5.8.6: fix greeting spam on restart - 30min grace, channel verification, file fallback

    Z User
    9ef0e78
  9. Fix: greeting spam on restart (persist to DB), crash stability, disconnect backoff, remove stealth references

    Z User
    162dea2

  1. feat: Add /chat and /test endpoints for direct AI testing

    Z User
    9e7536e
  2. fix: mineflayer-pathfinder ESM import (CJS compat)

    Z User
    1fcef6d
  3. v5.8.8: Gemma 3→4 upgrade + NODE_OPTIONS fix (same as zelin_deploy2)

    Z User
    ebd1c14
  4. v5.8.5d: Fix Pollinations API (model not found error), try multiple models

    Z User
    dd4a877
  5. v5.8.5c: Add node-llama-cpp for real local AI fallback, remove no-op shim

    Z User
    d626137
  6. v5.8.5: Gemma 4, MC Wiki, MC Player, anti-hallucination, CPU optimizations

    Z User
    ee826ee