assay-ai 1.5.1

Receipt-native AI safety toolkit

Assay

Verifiable evidence for AI systems. Independently verifiable, offline, without server access.

Logs record what you say happened. Assay makes the record tamper-evident, completeness-checkable, and independently verifiable -- including by someone who does not trust you. Two lines of code. Four exit codes.

pip install assay-ai && assay quickstart

Not this: Assay is not a logging framework, an observability dashboard, or a monitoring tool. It produces signed evidence bundles that a third party can verify offline. If you need Datadog, this isn't it.

See It -- Then Understand It

No API key needed. Runs on synthetic data:

assay demo-incident     # two-act scenario: honest PASS vs honest FAIL

Act 1: Agent uses gpt-4 with a guardian check. Integrity PASS, claims PASS. Act 2: Someone swaps the model and drops the guardian. Integrity PASS, claims FAIL.

That second result is an honest failure -- authentic evidence proving the run violated its declared standards. Not a cover-up. Exit code 1.

How that works

Assay separates two questions on purpose:

• Integrity: "Were these bytes tampered with after creation?" (signatures, hashes, required files)
• Claims: "Does this evidence satisfy our declared governance checks?" (receipt types, counts, field values)

Integrity	Claims	Exit	Meaning
PASS	PASS	0	Evidence checks out, behavior meets standards
PASS	FAIL	1	Honest failure: authentic evidence of a standards violation
FAIL	--	2	Tampered evidence
--	--	3	Bad input (missing files, invalid arguments)

The split is the point. Systems that can prove they failed honestly are more trustworthy than systems that always claim to pass.

Add to Your Project

# 1. Find uninstrumented LLM calls
assay scan . --report

# 2. Patch (one line per SDK, or auto-patch all)
assay patch .

# 3. Run + build a signed proof pack
assay run -c receipt_completeness -- python my_app.py

# 4. Verify
assay verify-pack ./proof_pack_*/

assay scan . --report finds every LLM call site (OpenAI, Anthropic, LangChain) and generates a self-contained HTML gap report. assay patch inserts the two-line integration. assay run wraps your command, collects receipts, and produces a signed 5-file proof pack. assay verify-pack checks integrity + claims and exits with one of the four codes above. Then run assay explain on any pack for a plain-English summary.

Why now: EU AI Act Articles 12 and 19 require logging and traceability for high-risk AI systems. SOC 2 CC7.2 requires evidence of monitoring. "We have logs on our server" is not independently verifiable evidence. Assay produces evidence that is.

CI Gate

Three commands, three exit codes, one lockfile:

assay run -c receipt_completeness -- python my_app.py
assay verify-pack ./proof_pack_*/ --lock assay.lock --require-claim-pass
assay diff ./baseline_pack/ ./proof_pack_*/ --gate-cost-pct 25 --gate-errors 0 --gate-strict

The lockfile catches config drift. Verify-pack catches tampering. Diff catches regressions and budget overruns. See Decision Escrow for the protocol model.

# Lock your verification contract
assay lock write --cards receipt_completeness -o assay.lock

Daily use after CI is green

Regression forensics:

assay diff ./proof_pack_*/ --against-previous --why

--against-previous auto-discovers the baseline pack. --why traces receipt chains to explain what regressed and which call sites caused it.

Cost/latency drift (from receipts):

assay analyze --history --since 7

Shows cost, latency percentiles, error rates, and per-model breakdowns from your local trace history.

Trust Model

What Assay proves, what it doesn't, and how to strengthen guarantees.

Assay detects:

• Retroactive tampering (edit one byte, verification fails)
• Selective omission under a completeness contract
• Claiming checks that were never run
• Policy drift from a locked baseline

Assay does not prevent:

• A fully fabricated false run (attacker controls the machine)
• Dishonest receipt content (receipts are self-attested)
• Timestamp fraud without an external time anchor

To strengthen guarantees:

• Transparency ledger (independent witness)
• CI-held org key + branch protection (separation of signer and committer)
• External timestamping (RFC 3161)

The cost of cheating scales with the complexity of the lie. Assay doesn't make fraud impossible -- it makes fraud expensive.

Commands

Command	Purpose
assay quickstart	One command: demo + scan + next steps
assay demo-incident	Two-act scenario: passing run vs failing run
assay demo-challenge	CTF-style good + tampered pack pair
assay demo-pack	Generate demo packs (no config needed)
assay onboard	Guided setup: doctor -> scan -> first run plan
assay scan	Find uninstrumented LLM call sites (--report for HTML)
assay patch	Auto-insert SDK integration patches into your entrypoint
assay run	Wrap command, collect receipts, build signed pack
assay verify-pack	Verify a Proof Pack (integrity + claims)
assay explain	Plain-English summary of a proof pack
assay analyze	Cost, latency, error breakdown from pack or --history
assay diff	Compare packs: claims, cost, latency (--against-previous, --why, --gate-*)
assay key list	List local signing keys and active signer
assay key rotate	Generate a new signer key and switch active signer
assay key set-active	Set active signing key for future runs
assay ci init github	Generate a GitHub Actions workflow
assay lock write	Freeze verification contract to lockfile
assay lock check	Validate lockfile against current card definitions
assay cards list	List built-in run cards and their claims
assay cards show	Show card details, claims, and parameters
assay mcp-proxy	Transparent MCP proxy: intercept tool calls, emit receipts
assay doctor	Preflight check: is Assay ready here?

Documentation

• Quickstart -- install, golden path, command reference
• Roadmap -- phases, product boundary, execution stack
• Decision Escrow -- protocol model: agent actions don't settle until verified
• For Compliance Teams -- what auditors see, evidence artifacts, framework alignment
• Repo Map -- what lives where across the Assay ecosystem
• Pilot Program -- early adopter program details

Scan Study

We scanned 30 popular open-source AI projects for tamper-evident audit trails. Found 202 high-confidence LLM SDK call sites across 21 projects. Zero had evidence emission at any call site. Full results.

Common Issues

• "No receipts emitted" after assay run: First, check whether your code has call sites: assay scan . -- if scan finds 0 sites, you may not be using a supported SDK yet. If scan finds sites, check: (1) Is # assay:patched in the file? Run assay scan . --report to see patch status per file. (2) Did you install the SDK extra (pip install assay-ai[openai])? (3) Did you use -- before your command (assay run -- python app.py)? Run assay doctor for a full diagnostic.

• LangChain projects: assay patch auto-instruments OpenAI and Anthropic SDKs but not LangChain (which uses callbacks, not monkey-patching). For LangChain, add AssayCallbackHandler() to your chain's callbacks parameter manually. See src/assay/integrations/langchain.py for the handler.

• assay run python app.py gives "No command provided": You need the -- separator: assay run -c receipt_completeness -- python app.py. Everything after -- is passed to the subprocess.

• Quickstart blocked on large directories: assay quickstart guards against scanning system directories (>10K Python files). Use --force to bypass: assay quickstart --force.

Get Involved

• Try it: pip install assay-ai && assay quickstart
• Questions / feedback: GitHub Discussions
• Bug reports: Issues
• Want this in your stack in 2 weeks? Pilot program -- we instrument your AI workflows, set up CI gates, and hand you a working evidence pipeline. Open a pilot inquiry.

Related Repos

Repo	Purpose
assay	Core CLI, SDK, conformance corpus (this repo)
assay-verify-action	GitHub Action for CI verification
assay-ledger	Public transparency ledger

License

Apache-2.0