tethral-acr 0.4.0

Python SDK for the ACR (Agent Composition Records) network. Register agents, log interactions, and query behavioral profiles through lenses: friction, coverage, stable corridors, failure registry, and trend. Raw signals, no synthetic labels.

tethral-acr

Python SDK for the ACR (Agent Composition Records) network.

Install

pip install tethral-acr

Quick Start

from tethral_acr import ACRClient

with ACRClient() as acr:
    # Register
    result = acr.register(
        public_key="your-unique-key-at-least-32-chars-long",
        provider_class="langchain",
    )
    agent_id = result["agent_id"]

    # Log an interaction
    acr.submit_receipt({
        "emitter": {"agent_id": agent_id, "provider_class": "langchain"},
        "target": {"system_id": "api:openai.com", "system_type": "api"},
        "interaction": {
            "category": "tool_call", "status": "success",
            "duration_ms": 800, "request_timestamp_ms": 1711978987442,
        },
        "anomaly": {"flagged": False},
    })

    # See what's costing you the most
    report = acr.get_friction_report(agent_id, scope="day")

API

Method	Description
register(public_key, provider_class, ...)	Register an agent
submit_receipt(receipt)	Submit a single receipt
submit_receipts(receipts)	Submit a batch (max 50)
update_composition(agent_id, ...)	Update skill composition
check_skill(skill_hash)	Check a skill before installing
check_agent(agent_id)	Look up an agent
get_system_health(system_id)	Get system health
get_active_signals()	Get skills with elevated anomaly signals
get_friction_report(agent_id, scope)	Friction analysis
health()	API health check

Data Collection

ACR collects interaction metadata only: target system names, timing, status, and provider class. No request/response content, API keys, prompts, or PII is collected. Full terms.

License

MIT