Skip to main content

LangChain tool adapters for Copass — drop-in discover/interpret/search tools for LangChain agents

Project description

copass-langchain

LangChain tool adapters for Copass. Python mirror of @copass/langchain. Pulls discover / interpret / search into any LangChain agent, and (optionally) wires them into LangGraph's ReAct prebuilt.

Install

pip install copass-langchain
# or, with the create_copass_agent convenience wrapper:
pip install copass-langchain[agent]

Depends on copass-core + copass-config + langchain-core. langgraph is optional — only needed for create_copass_agent.

Drop-in tools

from copass_core import ApiKeyAuth, CopassClient
from copass_langchain import copass_tools

client = CopassClient(auth=ApiKeyAuth(key="olk_..."))

tools = copass_tools(client=client, sandbox_id="sb_...")
# Pass `tools.all()` (a list of three StructuredTool instances) to your
# agent framework, or pull them individually:
# tools.discover, tools.interpret, tools.search

Tool descriptions and parameter descriptions come from copass-config — identical to the strings used across every other Copass adapter.

Window-aware retrieval (when a Context Window exists)

copass-core v0.1 does not yet ship a ContextWindow class (deferred to v0.2). Until then, copass_tools and CopassWindowCallback accept any object satisfying ContextWindowLike:

class ContextWindowLike(Protocol):
    def get_turns(self) -> list[ChatMessage]: ...
    def add_turn(self, turn: ChatMessage) -> Awaitable[None]: ...

When copass-core ships ContextWindow, it will satisfy this protocol and window-aware retrieval will light up without any consumer changes.

Full agent in one call

from copass_langchain import create_copass_agent
from langchain_anthropic import ChatAnthropic

agent = create_copass_agent(
    client=client,
    sandbox_id="sb_...",
    llm=ChatAnthropic(model="claude-opus-4-7"),
    # window=my_window,  # optional — enables window-aware retrieval + auto-mirroring
)

result = await agent.ainvoke({
    "messages": [("user", "why is checkout flaky?")]
})

Status

  • copass_tools — shipped.
  • CopassWindowCallback — shipped (generic on ContextWindowLike).
  • create_copass_agent — shipped (lazy-imports langgraph).
  • First-class ContextWindow integration — lands when copass-core v0.2 ships the primitive.

License

MIT.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

copass_langchain-0.1.0.tar.gz (8.8 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

copass_langchain-0.1.0-py3-none-any.whl (9.4 kB view details)

Uploaded Python 3

File details

Details for the file copass_langchain-0.1.0.tar.gz.

File metadata

  • Download URL: copass_langchain-0.1.0.tar.gz
  • Upload date:
  • Size: 8.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.10.9

File hashes

Hashes for copass_langchain-0.1.0.tar.gz
Algorithm Hash digest
SHA256 34210d0f1a06f7e69d6b7484777b06c3cbf638f1153d257e915ed63bc806ed37
MD5 bb1a4a4baeed79a31903201cc07ee342
BLAKE2b-256 cd837ccb8eefeee7fccf87f06cd101a86465ebdae479292ef2bcb39d0ff8355a

See more details on using hashes here.

File details

Details for the file copass_langchain-0.1.0-py3-none-any.whl.

File metadata

File hashes

Hashes for copass_langchain-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 1660aae571831c01236cee79c90e63ae777aa6a69067c80b8b05564fd02ec7bd
MD5 ca2c1d4f1c4351299dbc0f1643031efd
BLAKE2b-256 083a6e8d53823541bb416cdc457b59fdd5c50d8c3abd4ba35bda087a317cd4e9

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page