langchain - 💡(How to fix) Fix feat: add dependency-orchestrator utility package for runtime precondition gating on LangChain tools [1 comments, 2 participants]

Checked other resources

• This is a feature request, not a bug report or usage question.
• I added a clear and descriptive title that summarizes the feature request.
• I used the GitHub search to find a similar feature request and didn't find it.
• I checked the LangChain documentation and API reference to see if this feature already exists.
• This is not related to the langchain-community package.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-openrouter
• langchain-perplexity
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Feature Description

Summary Proposing a new utility package langchain-dependency-orchestrator under libs/dependency-orchestrator/ that enforces discovery-before-action as a runtime policy for LangChain tool pipelines.

Problem LangChain agents currently rely on prompt instructions to enforce tool ordering. This is fragile — models can skip prerequisite steps, leading to tool calls that fail silently or produce incorrect results due to missing context. Proposed Solution A wrap_tools() utility that wraps any list of BaseTool objects with precondition gating. When a tool is called without required context facts present, it automatically blocks or routes to the appropriate prerequisite tool instead of executing.

Scope New package only — no existing files modified No new external dependencies 55 unit tests included

Why this belongs in LangChain This is a reusable, framework-level concern that benefits any agent using multi-step tool pipelines where ordering and context correctness matter.

Use Case

When building LangChain agents that use multiple tools, the agent must often call a discovery or lookup tool before an action tool. For example, resolving an entity ID before updating a record. Currently, this ordering is enforced only through prompt instructions, which models can and do skip. This results in action tools being called with missing context, causing silent failures or incorrect results. There is no runtime mechanism to guarantee prerequisite tools have run and produced the required facts before a dependent tool is invoked.

Proposed Solution

A wrap_tools() utility that wraps any list of BaseTool objects with precondition gating at runtime. Each tool declares the context facts it requires, and before execution the wrapper checks whether those facts exist in a shared session store. If they don't, it automatically routes to the appropriate prerequisite tool instead of executing the dependent one. This requires no changes to existing LangChain internals, it works as a transparent layer on top of any BaseTool list and can be dropped into any existing agent with a single call.

Alternatives Considered

No response

Additional Context

No response