diff --git a/backend/docs/MEMORY_IMPROVEMENTS_SUMMARY.md b/backend/docs/MEMORY_IMPROVEMENTS_SUMMARY.md
index 3bb543af6..65449e6f3 100644
--- a/backend/docs/MEMORY_IMPROVEMENTS_SUMMARY.md
+++ b/backend/docs/MEMORY_IMPROVEMENTS_SUMMARY.md
@@ -33,6 +33,6 @@ No `current_context` argument is currently available in `main`.
 
 ## Verification Pointers
 
-- Implementation: `packages/harness/deerflow/agents/memory/prompt.py`
+- Implementation: `packages/harness/deerflow/agents/memory/backends/deermem/deermem/core/prompt.py`
 - Prompt assembly: `packages/harness/deerflow/agents/lead_agent/prompt.py`
 - Regression tests: `backend/tests/test_memory_prompt_injection.py`
diff --git a/backend/docs/RUN_INTERACTION_POLICY.md b/backend/docs/RUN_INTERACTION_POLICY.md
index 933576ec3..e672f57b8 100644
--- a/backend/docs/RUN_INTERACTION_POLICY.md
+++ b/backend/docs/RUN_INTERACTION_POLICY.md
@@ -3,7 +3,7 @@
 The harness resolves a `RunInteractionPolicy` from the run context and uses it
 as the single source for lead-agent tool visibility, clarification middleware
 behavior, sandbox network approval eligibility, and system-prompt guidance.
-The implementation lives in `deerflow/agents/interaction_policy.py`.
+The implementation lives in `../packages/harness/deerflow/agents/interaction_policy.py`.
 
 ## Modes and precedence
 
diff --git a/backend/docs/TITLE_GENERATION_IMPLEMENTATION.md b/backend/docs/TITLE_GENERATION_IMPLEMENTATION.md
index b0a7a0b03..1368c1c7d 100644
--- a/backend/docs/TITLE_GENERATION_IMPLEMENTATION.md
+++ b/backend/docs/TITLE_GENERATION_IMPLEMENTATION.md
@@ -42,7 +42,7 @@ title:
 
 ### 3. 文档
 
-#### [`docs/AUTO_TITLE_GENERATION.md`](../docs/AUTO_TITLE_GENERATION.md) (新建)
+#### [`docs/AUTO_TITLE_GENERATION.md`](AUTO_TITLE_GENERATION.md) (新建)
 - ✅ 完整的功能说明文档
 - ✅ 实现方式和架构设计
 - ✅ 配置说明
@@ -220,7 +220,7 @@ pytest
 
 ## 📚 相关资源
 
-- [完整文档](../docs/AUTO_TITLE_GENERATION.md)
+- [完整文档](AUTO_TITLE_GENERATION.md)
 - [LangGraph Middleware](https://langchain-ai.github.io/langgraph/concepts/middleware/)
 - [LangGraph State 管理](https://langchain-ai.github.io/langgraph/concepts/low_level/#state)
 - [LangGraph Checkpointer](https://langchain-ai.github.io/langgraph/concepts/persistence/)
diff --git a/backend/docs/checkpoint-retention-contract.md b/backend/docs/checkpoint-retention-contract.md
index 33d76845b..adcee94d0 100644
--- a/backend/docs/checkpoint-retention-contract.md
+++ b/backend/docs/checkpoint-retention-contract.md
@@ -94,7 +94,7 @@ resume, (c) branch from an older visible turn, and (d) orphan row counts.
   `parent_config` converts a cleanup into a thread-level outage (branch and
   regenerate fail loudly for every later turn).
 - Measurement first: proposals must include before/after numbers from
-  `scripts/benchmark/checkpoint/bench_channels.py` (per-thread rows/bytes,
+  `../scripts/benchmark/checkpoint/bench_channels.py` (per-thread rows/bytes,
   SQLite and Postgres) plus the contract test suite passing.
 
 ## History fast-path interaction (wiring requirement)
@@ -138,7 +138,7 @@ alist path without re-implementing the writer's merge semantics.
 reach state (preview + file reference under `.tool-results/`), so the
 "50 KB result re-snapshotted every step" scenario from the original report
 depends on which tools/paths bypass it. The probe
-(`scripts/benchmark/checkpoint/bench_tool_result_probe.py`) measures the
+(`../scripts/benchmark/checkpoint/bench_tool_result_probe.py`) measures the
 on-disk checkpoint delta for the wrapped vs unwrapped paths on the lead
 graph; subagent chains instantiate the same middleware by default. Any PR
 claiming a residual gap must name the concrete bypassing path and show its
diff --git a/backend/docs/plan_mode_usage.md b/backend/docs/plan_mode_usage.md
index 8caf30c93..9b82e677f 100644
--- a/backend/docs/plan_mode_usage.md
+++ b/backend/docs/plan_mode_usage.md
@@ -194,7 +194,7 @@ DeerFlow uses custom `system_prompt` and `tool_description` for the TodoListMidd
 - Comprehensive best practices section
 - Task completion requirements to prevent premature marking
 
-The custom prompts are defined in `_create_todo_list_middleware()` in `/Users/hetao/workspace/deer-flow/backend/packages/harness/deerflow/agents/lead_agent/agent.py:57`.
+The custom prompts are defined in `_create_todo_list_middleware()` in `../packages/harness/deerflow/agents/lead_agent/agent.py:57`.
 
 ## Notes