Skip to main content

AGLedger SDK — Accountability and audit infrastructure for agentic systems.

Project description

AGLedger Python SDK

The official Python SDK for AGLedger — accountability infrastructure for AI agents. Self-hosted. The Layer 3 accountability layer of the agent stack.

Learn more

  • agledger.ai — what AGLedger is and why Layer 3 accountability matters
  • How it works — the four-endpoint lifecycle: Record, Completion, Verdict, fulfill
  • Glossary — canonical definitions of Record, Completion, SCITT Receipt, Verdict, Settlement Signal
  • Documentation — installation, integration guides, API reference
  • Protocol (AOAP) — the coordination language behind AGLedger

Install

pip install agledger

Quick Start

import os
from agledger import AgledgerClient

client = AgledgerClient(
    api_key=os.environ["AGLEDGER_API_KEY"],
    base_url=os.environ["AGLEDGER_EXTERNAL_URL"],  # your AGLedger instance URL
)

# Create a Record. An agent key defaults the principal to itself; an admin
# key names the principal explicitly via principal_agent_id.
record = client.records.create(
    type="ACH-PROC-v1",
    contract_version="1",
    platform="internal",
    performer_agent_id="agt-123",
    criteria={"item_spec": "widgets", "quantity": {"target": 100}},
)

# Activate it
client.records.transition(record.id, "activate")

# Submit a completion
completion = client.completions.submit(
    record.id,
    evidence={"deliverable": "/out.pdf", "deliverable_type": "file_ref", "quantity_supplied": 95},
)

# Principal verdict
client.records.report_outcome(record.id, completion_id=completion.id, outcome="PASS")

Configuration

client = AgledgerClient(
    api_key="agl_agt_...",                              # or set AGLEDGER_API_KEY env var
    base_url="https://agledger.internal.example.com",   # your instance URL
    max_retries=2,                                      # default: 2
    timeout=30.0,                                       # default: 30s
)

Async Support

from agledger import AsyncAgledgerClient

async with AsyncAgledgerClient() as client:
    record = await client.records.get("rec-123")

Resources

records, completions (formerly receipts), verification, disputes, webhooks, reputation, events, schemas, compliance, health, admin (with admin.records + admin.vault sub-resources), a2a, agents, audit (with audit.org_reads_checkpoints and audit.vault_checkpoints), auth, capabilities, discovery, references, federation, federation_admin, verification_keys, scitt (SCITT/SCRAPI entries + Transparency Service keys), predicates (predicate schema discovery).

Offline Audit Export Verification

Verify a Record's hash-chained, Ed25519-signed audit export without calling the API:

from agledger.verify import verify_export

export_data = client.records.get_audit_export("rec-123")
result = verify_export(export_data.model_dump(by_alias=True))

if not result.valid:
    print(f"Broken at position {result.broken_at.position}: {result.broken_at.reason}")
# VerifyExportResult(valid=True, verified_entries=12, total_entries=12, ...)

Requires cbor2 (for COSE_Sign1 decoding) and cryptography (for Ed25519 verification):

pip install 'agledger[verify]'

Rewritten in 0.8.0 to decode canonical COSE_Sign1 envelopes (RFC 9052), walk the hash chain, and verify Ed25519 signatures. Format 2.0 (was 1.0 JCS + detached Ed25519). Pass public_keys={...} to override the export's embedded keys, or require_key_id="key-id" to reject exports signed by an unexpected key.

SCITT / SCRAPI

Register Signed Statements with the Transparency Service and retrieve Transparent Statements (Signed Statement + Receipt(s)):

receipt = client.scitt.entries.register(signed_statement)
# COSE_Sign1 Merkle inclusion proof per draft-ietf-cose-merkle-tree-proofs-18

transparent = client.scitt.entries.get(entry_id)
# Transparent Statement: Signed Statement with one or more Receipts embedded

keys = client.scitt.keys.list()
# COSE_KeySet of the Transparency Service's signing keys

Wire format is binary application/cose. Errors surface as RFC 9290 CBOR problem-details on APIError.raw_body.

Predicate Schemas

Fetch the canonical JSON Schemas for each predicate kind (record-state, settlement-signal, vault-checkpoint, schema-event, org-read, counter-attestation, federation-projection):

kinds = client.predicates.list()
schema = client.predicates.get("settlement-signal")

Attestation Export

Pull a Record's chain as a tagged COSE_Sign1 stream or a sigstore-bundle v0.3.2 projection for Rekor / in-toto / sigstore-policy-controller ingest:

cose_sequence = client.records.get_attestation(record_id)
# application/cose-sequence bytes (tagged COSE_Sign1 stream)

bundle = client.records.get_attestation_bundle(record_id)
# sigstore-bundle v0.3.2 projection

Vault Checkpoints

Per-record signed Merkle anchors are emitted every 6 hours, letting an auditor detect audit-vault TRUNCATE / DELETE tampering offline:

checkpoints = client.audit.vault_checkpoints.list(record_id="rec-123")

Licensing

AGLedger is free for single-node deployments (Docker Compose with bundled database). An Enterprise License is required for external database connections, federation, and multi-node deployments.

Full details: agledger.ai/pricing | License Agreement

SDK License

Proprietary. Copyright (c) 2026 AGLedger LLC. All rights reserved.

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

agledger-0.8.1.tar.gz (67.2 kB view details)

Uploaded Source

Built Distribution

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

agledger-0.8.1-py3-none-any.whl (61.9 kB view details)

Uploaded Python 3

File details

Details for the file agledger-0.8.1.tar.gz.

File metadata

  • Download URL: agledger-0.8.1.tar.gz
  • Upload date:
  • Size: 67.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.3

File hashes

Hashes for agledger-0.8.1.tar.gz
Algorithm Hash digest
SHA256 fb52cb1b41b61df8b9c29042043fa1be7a9c8e64553b251c5456605a5bf445ec
MD5 540bb38355df051712c54567f83b3669
BLAKE2b-256 b2b526656b4901456d7bb621b198a01bb74e78fdc0dbde7790704cbb2cc23943

See more details on using hashes here.

File details

Details for the file agledger-0.8.1-py3-none-any.whl.

File metadata

  • Download URL: agledger-0.8.1-py3-none-any.whl
  • Upload date:
  • Size: 61.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.3

File hashes

Hashes for agledger-0.8.1-py3-none-any.whl
Algorithm Hash digest
SHA256 2a5446342d253cfdb388e340c2f7d53ef383837c1977bb45f67d1e4a8d79ff7f
MD5 0866ab1991ffd217facd1098f1ca36a8
BLAKE2b-256 9f7eb35090b05a03f30028ad4a196cfc6ca13b052ec22fb42b4ec3f776d79860

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