avow 0.1.1

The trust kernel: one signed, offline-verifiable receipt envelope (avow) with a scoring face (assay) and an effect-gate face (writ).

avow

Proof of what your software decided — a receipt that is tamper-evident: edit it and it stops verifying.

---

Your card gets declined at a checkout.

You call the bank and ask why. Someone reads a reason off a screen: the fraud system scored the transaction as risky, so it blocked it.

But that reason is just a row in a database. Rows can be changed. Nobody — not you, not the bank's own auditor, not a regulator — can tell whether that number is what the software actually produced at the moment it blocked your card, or whether somebody adjusted it afterwards, once you complained.

That is the gap avow closes.

What it does

When your software makes a decision, avow has it write a receipt: a small record of exactly what was decided, sealed with cryptography at the moment of the decision.

Think of a store receipt — except this one cannot be reprinted or altered. You can hand it to anyone. They can check it on their own laptop, offline, with no access to your database, and get one of two answers:

• valid — this is exactly what the software decided, byte for byte
• invalid — someone changed it

There is no "close enough". Change one digit anywhere in it and the check fails.

Before you install

Two things that will otherwise trip you up:

• Python 3.13 or newer is required. On 3.12 or older pip install avow fails while resolving, with a message that does not mention the Python version. Check with python --version first.
• The package is avow; the command is assay. You pip install avow and import avow, but the command line the CLI extra installs is called assay. There is no avow command. (avow is the envelope everything is built on; assay is the scoring face that ships the CLI.)

See it catch a tampered record

python --version   # must be 3.13+
pip install avow

from pydantic import BaseModel, ConfigDict

from avow import generate_signing_key, public_key_hex, sign_payload, verify_signature


class FraudCheck(BaseModel):
    model_config = ConfigDict(frozen=True, extra="forbid")
    transaction_id: str
    decision: str
    risk_score: float
    model_version: str


key = generate_signing_key()        # the bank's private signing key
trusted_key = public_key_hex(key)   # published once; this is what checkers pin

receipt = sign_payload(
    FraudCheck(
        transaction_id="txn-9471",
        decision="blocked",
        risk_score=0.83,
        model_version="fraud-v4",
    ),
    key,
)

verify_signature(receipt, expected_public_key=trusted_key)
print("original receipt ..... VALID")

# Someone edits the stored record to make the block look better justified.
tampered = receipt.model_copy(
    update={"payload": receipt.payload.model_copy(update={"risk_score": 0.99})}
)

try:
    verify_signature(tampered, expected_public_key=trusted_key)
    print("edited receipt ....... VALID  <- this must never happen")
except Exception as exc:
    print(f"edited receipt ....... REJECTED ({type(exc).__name__})")

original receipt ..... VALID
edited receipt ....... REJECTED (ReplayMismatch)

Nudging 0.83 to 0.99 — a change that would be invisible in a database — makes the receipt fail to verify. That is the whole idea.

What a receipt proves, and what it does not

This is the part most signing libraries gloss over, so read it before you rely on avow.

A receipt proves integrity: the contents have not changed since they were signed.

It does not, on its own, prove authenticity — that your system is the one that signed it.

Here is why, concretely. The signer's public key travels inside the receipt, but outside the portion that is actually signed. So an attacker can write any payload they like, sign it with a key they generated themselves, and drop their own public key into the receipt. That forgery is internally consistent: its hash and its signature agree with each other perfectly.

The only thing that stops it is pinning — deciding in advance which public key you trust, obtaining it through a separate channel (the .pub file from keygen, your config, your secret manager), and passing that key to the verifier:

from pydantic import BaseModel, ConfigDict

from avow import generate_signing_key, public_key_hex, sign_payload, verify_signature


class FraudCheck(BaseModel):
    model_config = ConfigDict(frozen=True, extra="forbid")
    transaction_id: str
    decision: str
    risk_score: float
    model_version: str


bank = generate_signing_key()
trusted_key = public_key_hex(bank)      # what you pin, out-of-band

attacker = generate_signing_key()       # a key anyone can make in one line
forged = sign_payload(
    FraudCheck(
        transaction_id="txn-9471",
        decision="approved",            # a total fabrication
        risk_score=0.01,
        model_version="fraud-v4",
    ),
    attacker,
)

# The forgery is internally consistent: its hash and signature agree with each other.
print(f"forged receipt is self-consistent: {forged.payload_hash[:16]}... signed OK by attacker")

try:
    verify_signature(forged, expected_public_key=trusted_key)
    print("forged receipt ....... VALID  <- this must never happen")
except Exception as exc:
    print(f"forged receipt ....... REJECTED ({type(exc).__name__}: {exc.code})")

# The ONLY reason it was rejected is that we pinned the bank's key.
print(f"key inside forgery matches bank? {forged.public_key == trusted_key}")

forged receipt is self-consistent: sha256:441280ee2... signed OK by attacker
forged receipt ....... REJECTED (SignerMismatch: avow.signer_mismatch)
key inside forgery matches bank? False

Never trust the key embedded in the receipt. It rides along for convenience; it is not the trust anchor. verify_signature requires you to pass the key you already trust, precisely so this mistake is hard to make by accident.

Note the code: avow.signer_mismatch, not avow.signature_invalid. Those are two different events and they are coded apart, because you may want to react differently:

Code	Class	What happened
avow.signer_mismatch	SignerMismatch	Signed by a key you do not trust — a provenance failure. The signature is never even checked.
avow.signature_invalid	SignatureBytesInvalid	The signer matched, but the bytes fail the curve check — a tamper failure.

Both subclass SignatureInvalid, so except SignatureInvalid: still catches either one if you do not care which. You never have to match on the message text.

Three things you can sign

Avow ships as one installable package with three importable pieces. The first is the core; the other two are ready-made shapes built on it.

1. Anything — avow, the envelope

Shown above. You define what a decision looks like, avow seals and checks it. It never looks inside your data, so the same sign-and-verify code works for any record.

2. A measurement that refuses to overstate — assay

A number like "our model is 89% accurate" is only meaningful if enough examples stood behind it. Measure 12 cases and you can get any number you like; it is noise.

assay computes the score and its error bar, and when the sample is too thin it returns nothing at all rather than inventing a figure. Both outcomes come back inside a signed receipt.

pip install 'avow[assay]'

import random

from assay import score, verify
from assay.models import ScoreRequest
from assay.settings import AssaySettings
from avow import generate_signing_key, public_key_hex

key = generate_signing_key()
settings = AssaySettings()  # sample-size floor: 30


def evaluate(label: str, n: int) -> None:
    rng = random.Random(7)
    # a fraud model that is good, not perfect
    y_true = tuple(int(rng.random() < 0.3) for _ in range(n))
    y_score = tuple(min(1.0, max(0.0, rng.gauss(0.75 if t else 0.25, 0.22))) for t in y_true)

    receipt = score(
        ScoreRequest(metric="binary", metric_version="1", y_true=y_true, y_score=y_score),
        signing_key=key,
        settings=settings,
    )
    assert verify(receipt, expected_public_key=public_key_hex(key))

    r = receipt.payload
    if r.abstained:
        print(f"{label:<10} n={n:<4} accuracy = (none) -- {r.abstain_reason}")
    else:
        print(
            f"{label:<10} n={n:<4} accuracy = {r.score:.2f}  "
            f"95% interval [{r.interval_low:.2f}, {r.interval_high:.2f}]"
        )


evaluate("pilot", 12)
evaluate("full eval", 400)

pilot      n=12   accuracy = (none) -- assay.insufficient_samples
full eval  n=400  accuracy = 0.89  95% interval [0.86, 0.92]

The pilot run declines to produce a number. The full run reports 0.89 and admits the true value is somewhere in [0.86, 0.92]. The receipt also carries precision, recall, F1, PR-AUC, ROC-AUC, and a calibration report — see the reference section.

3. An action that was actually allowed — writ

Before your code does something irreversible — delete a record, move money, send an email — writ checks a policy. If the policy says no, the action never runs. Either way you get a signed receipt of what was asked and what was decided, so "the agent deleted it" and "we blocked the agent" are both provable after the fact.

This matters most when the caller is an AI agent you do not fully control.

from avow import content_hash, generate_signing_key, public_key_hex, verify_signature
from writ import Allowlist, EffectRequest, KeyholderEffector, governed_gate

key = generate_signing_key()
performed: list[str] = []  # stands in for the real system being changed


def perform(request: EffectRequest) -> None:
    """The privileged action. Reached ONLY through an allow decision."""
    performed.append(f"{request.action} {request.target}")


# The trusted host wires policy + action + key into the gate, then hands the agent
# exactly one thing: the gate. The agent never receives the key or the action itself.
agent_gate = governed_gate(
    Allowlist(frozenset({"read"})),
    KeyholderEffector(effect=perform, signing_key=key),
)

for action in ("read", "delete"):
    receipt = agent_gate(
        EffectRequest(
            action=action,
            target="customer-4471",
            args_digest=content_hash({"reason": "agent cleanup task"}),
        )
    )
    verify_signature(receipt, expected_public_key=public_key_hex(key))
    print(f"agent asked to {action:<6} -> {receipt.payload.decision:<5} (signed receipt verified)")

print(f"actually performed: {performed}")

agent asked to read   -> allow (signed receipt verified)
agent asked to delete -> deny  (signed receipt verified)
actually performed: ['read customer-4471']

The denied delete produced a signed receipt but never touched the system.

Command line

For signing and checking receipts without writing code. The distribution is avow; the command it installs is assay (there is no avow command). Python 3.13+ required.

pip install 'avow[cli]'                        # installs the `assay` command

assay --help                                   # note: `assay`, not `avow`
assay keygen --out signing.key                 # also writes signing.key.pub
echo '{"metric":"binary","metric_version":"1","y_true":[0,1,0,1],"y_score":[0.2,0.8,0.3,0.7]}' > req.json
assay score --request req.json --key signing.key --out receipt.json --ledger ledger.jsonl
assay verify --receipt receipt.json --public-key signing.key.pub

wrote signing key: signing.key
wrote public key: signing.key.pub
wrote receipt: receipt.json
OK: receipt verified

Auditing the ledger — verify-ledger

verify-ledger ships in the published avow 0.1.1 — pip install 'avow[cli]' (Python 3.13+; use a fresh venv) and run assay verify-ledger --help.

score also appended that receipt to ledger.jsonl. A ledger is checkable on its own — each entry is identified by the hash of its own contents, so an edit anywhere in the file is detectable without the signing key. Anyone holding the file can audit it:

assay verify-ledger --ledger ledger.jsonl

OK: ledger verified, 1 entry intact

Four samples is below the abstention floor, so that receipt honestly records "abstained":true and no score. Now edit the stored entry to claim a confident answer it never gave — change "abstained":true to "abstained":false, the sort of quiet correction that leaves no trace in an ordinary log — and ask again:

assay verify-ledger --ledger ledger.jsonl

FAIL: avow.ledger_integrity: tampered ledger entry: sha256:d82e3eda6264eb298565019b54ceb9f29d5d5ca7453ef49fda8904203cf7a6d4

Exit code 1, and the coded cause names both the failure and the entry that caused it. The check re-derives every entry's hash and fails closed on the first disagreement.

A ledger it cannot read is also a failure, not a pass. Mistype the path and you get:

FAIL: avow.ledger_unreadable: ledger is not a readable file: ledgr.jsonl

rather than OK: ledger verified, 0 entries intact — which would be a clean bill of health for a file that was never opened. The same applies to a directory in the file's place, a file whose permissions deny reading, and a line that is not a parseable receipt (avow.ledger_entry_malformed). A ledger that exists and is empty still passes with zero entries: that is a legitimate initial state, not an unanswered question.

Honest limits

Stated plainly, because each of these is a real boundary on what avow currently gives you.

• A receipt proves integrity, not authenticity, unless you pin the key. See the section above. This is the single easiest way to misuse the library.
• writ's enforcement is in-process (v0). The signing key and the privileged action live only inside the effector, which the gate captures in a closure; the agent receives the closure and never the effector, so the only route to the action is through the guard. But the credential is still in the same process, so same-process reflection (walking __closure__, for instance) could reach it. This is a capability-holding approximation, not true enforcement. Real un-bypassability — a separate-process broker or a WASM guest, where the caller's address space cannot reach the credential — is the v1 hardening. We claim no more than that.
• The v0 policy decider is a plain Python predicate (Allowlist). OPA/Rego is the v1 decider.
• writ signs the args_digest its caller hands it; it does not recompute it. The gate never sees the raw arguments, so it cannot check that the digest actually describes them. A caller that passes a digest of one thing and performs another gets a validly-signed receipt attesting the wrong arguments. What the receipt therefore proves is "this signer claimed this action, target and digest, and the policy decided this" — not "these are the arguments the effect ran with". Closing the gap means the request carrying the real arguments and the gate deriving the digest itself; that changes EffectRequest's public shape, so it is a v1 change, not a patch.
• Browser key custody is same-origin, not hardware-backed. In the browser build, keys are protected by the origin boundary alone — there is no secure element or OS keychain behind them.
• Receipts carry no timestamp. That is deliberate: it makes them reproducible (the same inputs always yield the same receipt). It also means a receipt cannot tell you when it was made. If you need that, record it outside the receipt or append to the ledger.

Reference

Packages and install matrix

One distribution, avow, exposes three import packages:

Package	What it is	Depends on	Install
avow	the shared envelope — sign, hash, verify a receipt	pydantic, pynacl, rfc8785	pip install avow
assay	the measurement face — an honest number in a receipt	avow + scikit-learn/scipy/numpy	pip install 'avow[assay]'
writ	the action face — a policy-gated effect, sealed as a receipt	avow only	pip install avow

Dependency arrows only ever point into avow: assay → avow and writ → avow. Avow imports neither, which is why installing the envelope alone never pulls in the heavy scientific stack. Importing assay without the [assay] extra raises a coded ScoringExtraMissing, not a raw ModuleNotFoundError.

How the sealing works

Some terms, each in one line:

• Content hash — a short fingerprint of some data. Change any byte and the fingerprint changes completely. Avow uses SHA-256.
• Canonicalization (RFC 8785 / JCS) — one fixed way to write a JSON object as bytes, so that the same data always produces the same bytes regardless of key order or language. Without it, two systems could hash "the same" record differently.
• Ed25519 — a signature scheme. A private key signs; the matching public key checks. Signing is deterministic: the same message and key always give the same signature.
• Frozen subject — the record being signed, declared immutable so it cannot be modified after signing.

sign_payload / verify_signature / payload_digest operate only on the canonical JSON of a frozen subject, and never inspect its fields. That is why the same envelope carries a measurement for assay and an action for writ with no change to the trust boundary.

Because payloads carry no timestamp, identical inputs yield an identical, reproducible, offline-verifiable receipt.

avow.ledger is an append-only, content-addressed JSONL log with a fail-closed integrity check, generic over the subject. Coded failures live in avow.errors (avow.* codes under AvowError).

Inside assay

A thin trust, honesty, and composition layer over reused libraries — it computes no metric math itself:

• scikit-learn for precision, recall, F1, PR-AUC, ROC-AUC and Brier score
• scipy.stats.bootstrap for percentile intervals, with a sample-size floor; below it, assay abstains rather than invent a point estimate
• population-weighted ECE (expected calibration error) for calibration
• a positive-weighted composite with a propagated interval

assay.receipt defines the measurement subjects; the envelope signs them. Errors are assay.* under AssayError. Every tunable — the sample floor, resample count, confidence level, bin count — lives in AssaySettings and is overridable via ASSAY_* environment variables.

Inside writ

writ.gate(request, policy, effector) evaluates a typed policy. On deny it seals a signed denial receipt and never runs the effect; on allow the effector runs the effect, then seals a signed effect receipt. Both outcomes are verifiable through the shared envelope.

EffectRequest.args_digest is a hash rather than the arguments themselves, so the signed record never carries raw payloads. It is the caller's claim about those arguments: the gate signs it without recomputing it.

See the honest limits above for exactly how far the enforcement seam and that digest go in v0.

Key custody and cross-language vectors

assay keygen (and avow.keys) write a 32-byte Ed25519 seed to a 0600 file and the public key to a companion .pub. Keys are never logged and never committed (*.key is gitignored). The public key also travels inside each receipt for convenience, but that embedded copy is not the trust anchor — a verifier pins the out-of-band key and passes it to verify.

testdata/vectors/ holds golden vectors generated by tests/gen_vectors.py (canonical bytes and hashes, plus receipts signed with a fixed non-secret test seed). The Python suite replays them in tests/test_vectors.py; the TypeScript @edgeproc/avow replays the same files byte for byte, so any RFC 8785 number-serialization divergence fails in CI rather than in production.

Working on avow itself

git clone https://github.com/hseshadr/assay.git && cd assay
uv sync --all-extras
uv run poe gate                          # Python: ruff, ruff-format, mypy --strict, xenon A, pytest
uv run poe gate-ts                       # TypeScript: biome, tsc --noEmit, vitest, build (needs pnpm)
uv run poe gate-all                      # both, mirroring CI's two jobs
uv run python demo/run_demo.py           # measurement face: 6 honesty acceptance cases
uv run python demo/unification_demo.py   # one envelope + one verifier, both faces

QUICKSTART.md is the shortest path from clone to a verified receipt.

See docs/ARCHITECTURE.md for the data-flow diagram, the import edges, and the native-vs-browser story.

Status

v0, deterministic — no LLM anywhere in the path.

Two gates, mirroring CI's two jobs. uv run poe gate covers Python only (ruff, ruff-format, mypy --strict, xenon A, pytest with statement and branch coverage against a floor); uv run poe gate-ts covers the TypeScript package (biome, tsc strict, vitest, build). uv run poe gate-all runs both.

Published releases: avow 0.1.1 on PyPI, @edgeproc/avow 0.1.1 on npm — see CHANGELOG.md for what each release contains. Read the honest limits above before depending on any of it.

License

MIT © Harish Seshadri