Skip to content

langgraph.md

Latest commit

 

History

History
242 lines (173 loc) · 12.4 KB

langgraph.md

File metadata and controls

242 lines (173 loc) · 12.4 KB

LangGraph: Guardrails and Fix Patterns

🧭 Quick Return to Map

You are in a sub-page of Agents & Orchestration.
To reorient, go back here:

Think of this page as a desk within a ward.
If you need the full triage and all prescriptions, return to the Emergency Room lobby.

Use this page when your orchestration uses LangGraph (graph nodes, tool nodes, channels, subgraphs) and you see tool loops, wrong snippets, role mixing, or answers that flip between runs. The table maps symptoms to exact WFGY fix pages and gives a minimal recipe you can paste.

Acceptance targets

  • ΔS(question, retrieved) ≤ 0.45
  • Coverage ≥ 0.70 to the intended section or record
  • λ stays convergent across 3 paraphrases and 2 seeds
  • E_resonance stays flat on long windows

Open these first

Typical breakpoints and the right fix

  • Tool node returns free text instead of strict JSON
    Fix: lock arguments, echo schema, and gate on validation.
    Open: Prompt Injection · Data Contracts

  • Edges create a loop or the run never reaches a terminal node
    Fix: add timeouts and BBCR bridge steps, log λ per hop, clamp variance with BBAM.
    Open: Logic Collapse

  • High similarity yet wrong meaning
    Fix: metric or index mismatch, or mixed write and read embeddings.
    Open: Embedding ≠ Semantic

  • Hybrid routes worse than a single retriever
    Fix: lock two stage query and add a deterministic reranker.
    Open: Query Parsing Split · Rerankers

  • Facts exist in the store yet never surface
    Fix: fragmentation or sharding misalignment.
    Open: Vectorstore Fragmentation

  • Citations missing or inconsistent across nodes
    Fix: require cite then explain and lock snippet fields.
    Open: Retrieval Traceability · Data Contracts

  • Long running graphs change style and drift logically
    Fix: split subgraphs and rejoin with BBCR, measure entropy and stop when it rises.
    Open: Context Drift · Entropy Collapse

  • Role confusion between planner and worker nodes
    Fix: stamp mem_rev and mem_hash, isolate namespaces, forbid cross section reuse.
    Open: Multi-Agent Problems

Fix in 60 seconds

  1. Measure ΔS
    Compute ΔS(question, retrieved) and ΔS(retrieved, expected anchor).
    Stable < 0.40, transitional 0.40 to 0.60, risk ≥ 0.60.

  2. Probe λ_observe
    Do a k sweep in retrieval. Reorder prompt headers. If λ flips, lock the schema and clamp with BBAM.

  3. Apply the module

  1. Verify
    Coverage ≥ 0.70. ΔS ≤ 0.45. Three paraphrases and two seeds with λ convergent.

Minimal LangGraph topology with WFGY checks

# Pseudocode: focus on control points
from langgraph.graph import StateGraph, END

def retrieve(state):
    # k sweep and unified analyzer across dense and sparse
    return {"context": retriever.search(state["q"], k=10)}

def assemble(state):
    # schema-locked prompt with cite first
    msg = prompt.format(context=state["context"], question=state["q"])
    return {"msg": msg}

def reason(state):
    ans = llm.invoke(state["msg"])
    return {"answer": ans}

def wfgy_checks(state):
    # compute ΔS(question, context) and enforce thresholds
    # stop when ΔS ≥ 0.60 or λ divergent
    metrics = metrics_and_trace(state["q"], state["context"], state["answer"])
    if metrics["risk"]:
        return {"halt": True, "metrics": metrics}
    return {"metrics": metrics}

g = StateGraph(dict)
g.add_node("retrieve", retrieve)
g.add_node("assemble", assemble)
g.add_node("reason", reason)
g.add_node("wfgy_checks", wfgy_checks)

g.add_edge("retrieve", "assemble")
g.add_edge("assemble", "reason")
g.add_edge("reason", "wfgy_checks")
g.add_edge("wfgy_checks", END)

app = g.compile()

What this enforces

  • Retrieval is observable and parameterized. Analyzer and metric are unified across paths.
  • Prompt is schema locked with cite first. Snippet fields are required.
  • Post generation WFGY checks can halt the run when ΔS is high or λ flips.

Specs and recipes RAG Architecture & Recovery · Retrieval Playbook · Retrieval Traceability · Data Contracts

LangGraph specific gotchas

  • Mixed embedding functions between write and read. Rebuild with explicit metric and normalization. See Embedding ≠ Semantic

  • Event storms caused by branching edges. Add idempotency keys and dedupe on {source_id, mem_rev, index_hash}. See Retrieval Traceability

  • Subgraph recursion without a guard. Cap depth and time, insert BBCR bridge steps. See Logic Collapse

  • Memory modules re assert old facts after refresh. Stamp mem_rev and mem_hash. See pattern_memory_desync

  • Streams are merged without source fences. Add per source headers and forbid cross section reuse. See pattern_symbolic_constraint_unlock

When to escalate

  • ΔS remains ≥ 0.60 Rebuild the index using the checklists and verify with a small gold set. Retrieval Playbook

  • Identical input yields different answers across runs Check version skew and session state. Pre-Deploy Collapse

🔗 Quick-Start Downloads (60 sec)

Tool Link 3-Step Setup
WFGY 1.0 PDF Engine Paper 1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + <your question>”
TXT OS (plain-text OS) TXTOS.txt 1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” — OS boots instantly

Explore More

Layer Page What it’s for
⭐ Proof WFGY Recognition Map External citations, integrations, and ecosystem proof
⚙️ Engine WFGY 1.0 Original PDF tension engine and early logic sketch (legacy reference)
⚙️ Engine WFGY 2.0 Production tension kernel for RAG and agent systems
⚙️ Engine WFGY 3.0 TXT based Singularity tension engine (131 S class set)
🗺️ Map Problem Map 1.0 Flagship 16 problem RAG failure taxonomy and fix map
🗺️ Map Problem Map 2.0 Global Debug Card for RAG and agent pipeline diagnosis
🗺️ Map Problem Map 3.0 Global AI troubleshooting atlas and failure pattern map
🧰 App TXT OS .txt semantic OS with fast bootstrap
🧰 App Blah Blah Blah Abstract and paradox Q&A built on TXT OS
🧰 App Blur Blur Blur Text to image generation with semantic control
🏡 Onboarding Starter Village Guided entry point for new users

If this repository helped, starring it improves discovery so more builders can find the docs and tools.
GitHub Repo stars

要我接著做哪一個？建議順序：crewai.md → smolagents.md → rewind_agents.md（若你有其他優先順序，直接說 GO + 檔名）。