diff --git a/env.example b/env.example
index ac6fb40e8d..db41c7df34 100644
--- a/env.example
+++ b/env.example
@@ -191,6 +191,21 @@ SUMMARY_LANGUAGE=English
 ### Entity types that the LLM will attempt to recognize
 # ENTITY_TYPES='["Person", "Creature", "Organization", "Location", "Event", "Concept", "Method", "Content", "Data", "Artifact", "NaturalObject"]'
 
+### Prompt overrides (optional, env vars)
+### Override extraction/query prompts without modifying source code.
+### String prompts: set the env var to the full prompt text.
+### List prompts (examples): set the env var to a JSON array of strings.
+### If not set, built-in defaults are used. See lightrag/prompt.py for defaults.
+### Note: this is a lightweight bridge until the full prompt template system (#2623) lands.
+# ENTITY_EXTRACTION_SYSTEM_PROMPT=<your custom extraction system prompt>
+# ENTITY_EXTRACTION_USER_PROMPT=<your custom extraction user prompt>
+# ENTITY_EXTRACTION_EXAMPLES=<JSON array of example strings>
+# SUMMARIZE_ENTITY_DESCRIPTIONS_PROMPT=<your custom summarization prompt>
+# RAG_RESPONSE_PROMPT=<your custom RAG response prompt>
+# NAIVE_RAG_RESPONSE_PROMPT=<your custom naive RAG response prompt>
+# KEYWORDS_EXTRACTION_PROMPT=<your custom keywords extraction prompt>
+# KEYWORDS_EXTRACTION_EXAMPLES=<JSON array of example strings>
+
 ### Chunk size for document splitting, 500~1500 is recommended
 # CHUNK_SIZE=1200
 # CHUNK_OVERLAP_SIZE=100
diff --git a/lightrag/prompt.py b/lightrag/prompt.py
index dcd829d487..ed318f6992 100644
--- a/lightrag/prompt.py
+++ b/lightrag/prompt.py
@@ -1,4 +1,6 @@
 from __future__ import annotations
+import json
+import os
 from typing import Any
 
 
@@ -430,3 +432,43 @@
 
 """,
 ]
+
+# ── Environment variable overrides ───────────────────────────────────
+# Allow users to customize prompts via environment variables without
+# modifying this file. Each override is optional — if the env var is
+# not set, the hardcoded default above is used.
+#
+# This is a lightweight bridge until the full prompt template management
+# system (see #2623 / #2652) is available.
+#
+# String prompts: set the env var to the full prompt text.
+# List prompts (examples): set the env var to a JSON array of strings.
+
+_PROMPT_ENV_OVERRIDES: dict[str, str] = {
+    "entity_extraction_system_prompt": "ENTITY_EXTRACTION_SYSTEM_PROMPT",
+    "entity_extraction_user_prompt": "ENTITY_EXTRACTION_USER_PROMPT",
+    "summarize_entity_descriptions": "SUMMARIZE_ENTITY_DESCRIPTIONS_PROMPT",
+    "rag_response": "RAG_RESPONSE_PROMPT",
+    "naive_rag_response": "NAIVE_RAG_RESPONSE_PROMPT",
+    "keywords_extraction": "KEYWORDS_EXTRACTION_PROMPT",
+}
+
+_LIST_PROMPT_ENV_OVERRIDES: dict[str, str] = {
+    "entity_extraction_examples": "ENTITY_EXTRACTION_EXAMPLES",
+    "keywords_extraction_examples": "KEYWORDS_EXTRACTION_EXAMPLES",
+}
+
+for _key, _env_var in _PROMPT_ENV_OVERRIDES.items():
+    _val = os.environ.get(_env_var)
+    if _val is not None:
+        PROMPTS[_key] = _val
+
+for _key, _env_var in _LIST_PROMPT_ENV_OVERRIDES.items():
+    _val = os.environ.get(_env_var)
+    if _val is not None:
+        try:
+            parsed = json.loads(_val)
+            if isinstance(parsed, list):
+                PROMPTS[_key] = parsed
+        except (json.JSONDecodeError, TypeError):
+            pass  # Ignore malformed JSON — keep the default