Issues

• [Feature Request]: Pluggable decision models (Jev & alikes) for entity matching and retrieval

  enhancement
  HKUDS/LightRAG#4129 · Filter by author semidark opened on Sep 30, 2026

• Transition WebUI to account-only authentication before role-based authorization

  refacttrackedui
  HKUDS/LightRAG#4125 · Filter by author danielaskdd opened on Sep 29, 2026

• [Bug]: POST /graph/relation/edit and DELETE /graph/relation/delete refuse the same raw spellings the create endpoints accepted

  HKUDS/LightRAG#4114 · Filter by author passionworkeer opened on Sep 27, 2026

• [Bug]: POST /graphs/relation rejects entity names that POST /graphs/entity accepted

  HKUDS/LightRAG#4113 · Filter by author passionworkeer opened on Sep 27, 2026

• [Bug]: F/R source-span provenance is lost for consecutive token windows on emoji-heavy documents (up to 55% of the text at default chunk settings)

  HKUDS/LightRAG#4112 · Filter by author passionworkeer opened on Sep 27, 2026
  2

• What goes in LLM_BINDING_API_KEY when the host has no key?

  HKUDS/LightRAG#4106 · Filter by author TheCheetah11 opened on Sep 26, 2026

• [Bug]: audit_kg_integrity resolves provenance from the capped source_id, not chunk tracking

  HKUDS/LightRAG#4083 · Filter by author JessYanCoding opened on Sep 24, 2026

• [Feature Request]: Allow clearing only the query/keywords answer cache, independent of document deletion

  enhancement
  HKUDS/LightRAG#4082 · Filter by author AdalbertoDam opened on Sep 24, 2026

• [Bug]: bedrock_complete ignores LLM_TIMEOUT, so GPT-6 Astra extraction calls hit botocore's 60 s read timeout

  HKUDS/LightRAG#4075 · Filter by author kimnamu opened on Sep 24, 2026

• [Feature Request]: UUID-based config storage anchoring and offline migration

  enhancement
  HKUDS/LightRAG#4059 · Filter by author danielaskdd opened on Sep 23, 2026

• feat(tools): add bounded count-aware VDB consistency checks

  enhancementtracked
  HKUDS/LightRAG#4058 · Filter by author danielaskdd opened on Sep 23, 2026

• LoLLMs HTTP errors are returned as successful generation output

  HKUDS/LightRAG#4056 · Filter by author yeliheng-010 opened on Sep 23, 2026

• [Feature Request]: Chunk-level add / delete / modify for processed documents

  HKUDS/LightRAG#4055 · Filter by author itsskofficial opened on Sep 23, 2026

• CI's offline gate never runs 1,589 tests, and four of them have rotted

  HKUDS/LightRAG#4033 · Filter by author YHC66 opened on Sep 21, 2026
  1

• [Bug]: hardcoded response_format={"type": "json_object"} in keyword extraction breaks OpenAI-compatible servers that only accept json_schema/text (e.g. LM Studio) — hybrid/local/global queries always fail

  HKUDS/LightRAG#4023 · Filter by author moracabanas opened on Sep 20, 2026

• Configuration storage as its own category: a config_dir and fixed container names, retiring the reserved workspace

  refacttracked
  HKUDS/LightRAG#4020 · Filter by author danielaskdd opened on Sep 19, 2026

• RedisKVStorage acquires its shared connection-pool reference at construction, so a failed LightRAG construction leaks it

  bugtrackedtriage
  HKUDS/LightRAG#4016 · Filter by author danielaskdd opened on Sep 18, 2026
  1

• Configuration storage: one workspace-independent place for server and per-workspace settings

  HKUDS/LightRAG#4006 · Filter by author danielaskdd opened on Sep 18, 2026

• [Feature Request]: per-call provider context — no supported way to attach request-scoped data to a query's LLM, embedding, or rerank calls

  HKUDS/LightRAG#4003 · Filter by author pennycoders opened on Sep 17, 2026

• Vector storage: a model or dimension change must fail closed, and lightrag-rebuild-vdb must be able to recover on every backend

  HKUDS/LightRAG#3978 · Filter by author danielaskdd opened on Sep 16, 2026
  2

• clear_documents leaves other embedding models' vector containers intact

  trackedtriage
  HKUDS/LightRAG#3974 · Filter by author danielaskdd opened on Sep 16, 2026
  1

• _generate_collection_suffix folds distinct model names onto one container name

  HKUDS/LightRAG#3967 · Filter by author danielaskdd opened on Sep 15, 2026

• LightRAG evaluation with reproduce gives me very different results with local LLMs

  question
  HKUDS/LightRAG#3931 · Filter by author s-osimi-univpm opened on Sep 13, 2026

• [Feature Request]:Expose graph extraction and query APIs as stable public interface

  enhancement
  HKUDS/LightRAG#3919 · Filter by author LXXiaogege opened on Sep 11, 2026

• ainsert_custom_chunks: make custom-chunk documents self-describing and reproducible (follow-up to #3400)

  backendenhancementtracked
  HKUDS/LightRAG#3873 · Filter by author danielaskdd opened on Sep 8, 2026