Skip to main content

smolagents-aer1

Every tool call your smolagents agent makes, recorded as an AER-1 verifiable execution receipt. Five lines of code. No new infrastructure.

AER-1 is an open IETF Internet-Draft (draft-zambo-aer1) defining a small vocabulary for recording one AI agent tool call as a portable, independently checkable receipt. A receipt proves the recorded result was not changed. It does not prove the tool was correct.

For whole-run workflow receipts with a Merkle root (AER-1 Section 8), see aer1-smolagents.

Install

pip install smolagents-aer1

10-minute quickstart

from smolagents import CodeAgent, InferenceClientModel
from smolagents_aer1 import AER1Recorder

recorder = AER1Recorder()
agent = CodeAgent(
    tools=[],
    model=InferenceClientModel(),
    step_callbacks=[recorder],
)
agent.run("What is 17 * 24?")

print(f"{len(recorder.receipts)} receipts recorded")
recorder.save("receipts.jsonl")

That is the whole integration. step_callbacks is smolagents' blessed hook for observing steps (the same pattern tracing integrations use). The recorder watches each action step and emits one receipt per tool call.

Chain it: a job, not a lookup

One call answers a question. A job chains calls and ends in a record a person can open. Same agent, longer instruction:

agent.run("Check the live price of BTC, ETH and SOL, save the run to memory, then wrap it into a handover addressed to Dana Whitfield, with the receipt link for every call.")

The recorder receipts every call in the chain; when the agent's tools are Zambo's, the wrap is the handover call at the end of the run.

Three prices, a save and the wrap fit inside one free day: every tool free once a day, plus 3 continue calls, and one free wrap-up handover per day. A longer run finishes in one sitting on a Day Pass ($0.99, unlimited for 24 hours). The handover is the part the person opens: the steps, the saved results, and the receipt for every call, checkable instead of trusted. Ready-made run cards for jobs shaped like this: https://rambozambodotdev.gitlab.io/aer1-hub/handover-templates.html

What a receipt looks like

{
  "id": "3f9a2c1e-7b4d-4f8a-9c2e-1a5b6d7e8f90",
  "receipt_schema_version": "0.3",
  "created_at": "2026-10-05T21:20:00.123456Z",
  "tool": {"name": "python_interpreter", "version": "1.26.0", "scope": "public"},
  "provenance_class": "EXECUTED BY SMOLAGENTS",
  "canonical_bytes": "eyJpbnB1dHMiOi...",
  "output_hash": "sha256:9f2c...",
  "verification_status": "verified"
}

The canonical bytes are a deterministic JSON payload (tool, inputs, output). Anyone can base64-decode them, recompute SHA-256, and compare with output_hash. That is the entire verification procedure.

Verify a receipt yourself

from smolagents_aer1 import verify_receipt

failures = verify_receipt(recorder.receipts[0])
assert failures == [], failures
print("receipt checks out")

Or run the full open conformance suite against your receipts: https://gitlab.com/rambozambodotdev/zambo

Notes

  • Provenance class is EXECUTED BY SMOLAGENTS: the recorder runs inside the agent process and records what the agent did.
  • CodeAgent records each code execution as one python_interpreter tool call. ToolCallingAgent records each named tool call separately.
  • When one step contains several tool calls, each receipt commits to the step's full observation string.
  • The tool.version field records the smolagents runtime version.
  • Receipts are local JSON. Nothing is sent anywhere. Publishing or anchoring them is your choice.

License

MIT. See LICENSE.

Where these receipts run live: Zambo

smolagents-aer1 gives smolagents agents a verifiable AER-1 receipt for every tool call. The live home of those receipts is Zambo, the verifiable work layer for AI (live counts: https://zambo.dev/numbers/). Connect any AI to the live endpoint at https://zambo.dev/mcp or add Zambo to your AI in one step at https://zambo.dev/add/

Free tier: every tool free once a day, plus 3 continue calls. One free wrap-up handover per day, and pickup is always free. The Day Pass is $0.99 for 24 hours, unlimited.

Install:

pip install smolagents-aer1

Zambo, built by Brennan Zambo, the creator of the AER-1 standard, runs 153 tools in total, 37 on the default surface, behind one MCP endpoint: https://zambo.dev/mcp.

Metadata

Release files for smolagents-aer1 0.1.5

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for smolagents-aer1 0.1.5
File Size Uploaded Provenance Actions
smolagents_aer1-0.1.5.tar.gz 7.1 kB Without provenance Details for smolagents_aer1-0.1.5.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for smolagents-aer1 0.1.5
File Interpreter ABI Platform Actions
smolagents_aer1-0.1.5-py3-none-any.whl Python 3 none any Details for smolagents_aer1-0.1.5-py3-none-any.whl

Total release size: 13.6 kB

Release files / smolagents_aer1-0.1.5.tar.gz

Download URL smolagents_aer1-0.1.5.tar.gz
Size 7.1 kB
Tags Source
SHA-256 checksum
How to use checksums
af75c8177fa7c5372bbfa399b890420cafc6501bf206bbb7c45cdabfcff18928
BLAKE2b-256 checksum
How to use checksums
9c205b71cc4a0c537c4c5177854d3097c21b25088a7fc10b7ec22c23a3da53a2
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via rambo-pypi-grab/0.1.0

Release files / smolagents_aer1-0.1.5-py3-none-any.whl

Download URL smolagents_aer1-0.1.5-py3-none-any.whl
Size 6.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
ac66873e9655b77a982ef6fc75f7df41feedc8d8fb412b62da4affb7b0d7211c
BLAKE2b-256 checksum
How to use checksums
7ea33b8526570dced9db132171bed4d34706cc02d6493174eeff53ca6855bf82
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via rambo-pypi-grab/0.1.0

Security for smolagents-aer1 0.1.5

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page