entire-adapter 0.1.0

LangGraph and CrewAI adapter for Entire checkpoint tracking.

Entire Adapter

Python adapter that lets LangGraph/LangChain and CrewAI agents participate in Entire checkpoint tracking.

The package provides:

• EntireCallbackHandler for LangChain and LangGraph callback configs.
• EntireCrewAIListener for CrewAI's event bus.
• entire-agent-entire-adapter, an Entire external-agent protocol executable discovered as entire-adapter.

Why

Custom Python agents often edit files through tools, but their prompt/tool context is easy to lose once the final Git diff lands. Entire Adapter bridges those frameworks into Entire's lifecycle hooks so commits can be linked to the agent session and checkpoint history.

Install

For LangGraph/LangChain:

pip install "entire-adapter[langgraph]"

For CrewAI:

pip install "entire-adapter[crewai]"

For local development from this repo:

pip install -e ".[langgraph,crewai,dev]"

Enable the adapter inside a Git repo where Entire is installed:

entire enable --agent entire-adapter --telemetry=false
entire agent list

The adapter does not auto-commit your work. Persistent checkpoint metadata is finalized by Entire's existing Git hooks when you commit.

LangGraph / LangChain

from entire_adapter import EntireCallbackHandler

entire = EntireCallbackHandler(agent_label="reviewer")

result = graph.invoke(
    {"prompt": "Refactor the parser"},
    config={"callbacks": [entire]},
)

print(entire.session_id)

The handler maps:

• top-level on_chain_start to session-start and turn-start
• on_agent_action and on_llm_end to transcript reasoning/text records
• on_tool_start to transcript tool context
• on_tool_end and on_tool_error to transcript tool result and turn-end
• top-level on_chain_end and on_chain_error to session-end

See usage_example.py for a minimal graph example.

CrewAI

Create and keep a listener instance alive in the module where your crew or flow runs:

from entire_adapter import EntireCrewAIListener

entire_listener = EntireCrewAIListener(agent_label="research-crew")

The listener registers handlers for crew kickoff, agent execution, and tool usage events. Tool completions trigger Entire turn-end hooks so file changes can become checkpoints.

External-Agent Protocol

Entire discovers external agents from executables named:

entire-agent-<name>

This package installs:

entire-agent-entire-adapter

So the Entire agent identity is:

entire-adapter

Protocol commands include:

entire-agent-entire-adapter info
entire-agent-entire-adapter detect
entire-agent-entire-adapter parse-hook --hook turn-end < payload.json
entire-agent-entire-adapter read-transcript --session-ref .git/entire-adapter/sessions/demo.jsonl
entire-agent-entire-adapter compact-transcript --session-ref .git/entire-adapter/sessions/demo.jsonl

Metadata And Reasoning

The adapter writes an append-only JSONL transcript outside the worktree when possible:

.git/entire-adapter/sessions/<session-id>.jsonl

Each record includes:

• framework
• agent_label
• session_id
• callback/event metadata
• prompt text
• tool name
• tool input
• tool output or error
• assistant/LLM text when available

Entire reads this transcript through the external-agent protocol so reasoning context can be displayed alongside file changes.

Agent Identity

Entire sees one external agent:

entire-adapter

Distinct user agents are separated by generated session IDs and metadata:

langgraph-reviewer-77f263df4f684baf
crewai-research-crew-77f263df4f684baf

Use agent_label to identify the specific agent or workflow in your project.

Graceful Degradation

By default, the adapter never crashes your agent if Entire is unavailable. It logs one warning and continues when:

• entire is not installed or not on PATH
• the current directory is not an Entire-enabled Git repo
• an Entire hook command exits non-zero
• a hook call times out

Use strict=True if you want those failures to raise:

EntireCallbackHandler(strict=True)
EntireCrewAIListener(strict=True)

Verification

Run tests:

python -m pytest -q

Build and check the package:

python -m build
python -m twine check dist/*

Manual live checkpoint test commands are in adapter_live_test/README.md.

After running an agent that edits files:

git status
git add .
git commit -m "test entire adapter"
entire checkpoint list --session "$SESSION_ID"

Use the real session ID printed by the adapter, not an angle-bracket placeholder.

Live MVP Result

A local live test validated the full path:

LangGraph callback -> Entire Adapter -> Entire hook -> Git commit linkage -> Entire checkpoint list

The successful checkpoint output included:

branch       master
session      langgraph-live-test-77f263df4f684baf
checkpoints  1

● e23a3f26c8c4  "Create a visible file change for Entire checkpoint testing."

Development

python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -e ".[langgraph,crewai,dev]"
python -m pytest -q

Release notes are in CHANGELOG.md. Architecture/interview notes are in docs/INTERVIEW_BRIEF.md.