agent-assure

Output equivalence is not process equivalence.

Same approval. Missing evidence link. Configured CI gate blocked.

agent-assure catches regressions in declared, structured process evidence that final-answer-only checks can miss. It turns privacy-filtered run records into reproducible comparisons, reviewer-facing artifacts, portable evidence packets, and ordinary CI gate signals. Evidence strength follows its recorded origin: fixture values are authored test inputs, instrumented-adapter values remain producer-attested, and direct model responses do not independently prove that a tool, evidence lookup, policy check, or human review occurred.

Local-first · offline flagship demo · versioned artifacts · CI-native · no hosted control plane required

Run the offline demo · Inspect the reviewer output

For AI leaders · For architects · For engineers

10 deterministic fixture cases · 0 decision fields changed · claim-duration: linked → missing · classification: new_failure · configured CI gate: blocked

Note

This is a bundled deterministic fixture demonstration, not a model benchmark, live-model result, or customer outcome. It requires no provider API key, network call, or token spend.

Read the flagship demonstration

Choose your path

Your role	Start with the question that matters
Chief AI Officers and release owners	Did a release preserve its declared controls? See the local evidence and portable handoff for repeatable review. Read the AI leader brief.
AI/ML architects and platform owners	Where does it fit, what crosses the trust boundary, and which contracts are stable? Review the architecture · Inspect the API surface.
AI/ML engineers	How do I turn observable expectations and privacy-filtered run records into a configured CI decision? Follow the engineering guide.

What it can surface

A final-answer-only check can see no decision-field change while a declared release expectation regresses:

• Evidence and RAG support: a required source or material claim-to-evidence link disappears, or declared corpus and retrieval identity changes. Example: MATERIAL_CLAIM_MISSING_EVIDENCE → new_failure. The development surface also includes an authority-scoped, deterministic controlled evidence-sensitivity contract. Its v1 producer is a declarative synthetic fixture harness, not evidence that a model used contextual evidence instead of parametric memory. A separate repeated paired protocol supports the same narrow binary response relation for prebound stochastic live studies. Its fixed planned-frame composite rate includes every frozen cluster, pessimistically scores a non-analyzable cluster as zero, and reports observed/analyzable counts separately. Verdict-bearing packets and graphs bind the exact candidate RunSet, while both source execution configurations remain anchored to their predeclared arms. Conclusions remain conditional on statistical sufficiency, exact subject binding, and declared cluster assumptions. The untagged development surface also provides a preregistered real-model study workflow, but this repository contains no real-provider study result.
• Human review: a required route or performed-review record is missing.
• Provider, tool, and privacy boundaries: a forbidden provider or tool appears, or declared route, redaction state, or detector identity changes.
• Usage and reliability: retries, tool calls, tokens, latency, rate-limit events, or declared estimated cost change materially.
• Streaming integrity: events are replayed, duplicated, conflicting, or outside the declared sequence contract.
• Protocol-bound live behavior: repeated observations drift outside a declared protocol or comparison boundary, or a paired evidence-response study is underpowered, structurally invalid, or inconsistent with its exact pre-execution arm bindings.

A surfaced difference may be blocking, review-only, or informational. Usage and reliability deltas block only when a suite or policy declares that behavior.

Quickstart

Requires Python 3.11 or newer.

pip install agent-assure
agent-assure demo flagship --out .tmp/demo/flagship --clean

The installed package runs the bundled deterministic fixture from any directory—no repository clone, provider API key, network call, or token spend.

output equivalence: preserved
missing evidence link: claim-duration
reason code: MATERIAL_CLAIM_MISSING_EVIDENCE
classification: new_failure
CI gate: blocked as expected

The demo wrapper exits 0 only when it verifies that the expected regression was caught. The underlying candidate evaluation, comparison, and CI commands remain strict and exit nonzero for the blocking finding.

Actual reviewer output

_{Screenshot of the reviewer-facing evidence-diff.html produced from the same bundled fixture. Open the image to inspect it at full resolution.}

Reviewer-facing artifacts

Key artifacts are written under .tmp/demo/flagship:

Artifact	Review purpose
demo-summary.json	Machine-readable demonstration result
baseline-report/evaluation-summary.json	Baseline behavior against declared expectations
comparison-report/comparison-summary.json	Controlled baseline-to-candidate classification
ci-report/evidence-packet.json	Portable machine-readable review handoff
evidence-diff.html	Self-contained human-readable evidence diff

How this README evidence view is verified against the fixtures

Flagship regression at a glance

This diagram is checked in CI against the bundled flagship fixtures, keeping README claims aligned with the evidence produced by the project itself.

flowchart LR
    subgraph OutputCheck["Ordinary visible-output check"]
        BOut["Baseline output<br/>recommendation=approve<br/>outcome=approve"]
        COut["Candidate output<br/>recommendation=approve<br/>outcome=approve"]
        Same["Visible answer unchanged"]
        BOut --> Same
        COut --> Same
    end

    subgraph InvariantCheck["agent-assure invariant check"]
        BEv["Baseline evidence<br/>claim-duration linked"]
        CEv["Candidate evidence<br/>claim-duration missing link"]
        Pass["Baseline evaluation: pass"]
        Fail["Candidate evaluation: fail<br/>MATERIAL_CLAIM_MISSING_EVIDENCE"]
        BEv --> Pass
        CEv --> Fail
    end

    Same --> Tension["Output unchanged<br/>but governance invariant regressed"]
    Equiv["Fixture equivalence: pass"] --> Compare["Baseline-to-candidate comparison"]
    Pass --> Compare
    Fail --> Compare
    Tension --> Compare

    Compare --> NewFailure["Classification: new_failure"]

Loading

Where it fits

agent-assure complements the evaluation, observability, runtime-control, and governance systems teams already use.

Layer	Primary question	Relationship to agent-assure
Output and agent evals	Does the answer, trajectory, tool use, or component meet its quality criteria?	Adds source-aware checks for declared process-evidence expectations.
Observability and tracing	What happened during execution?	Consumes versioned, privacy-filtered evidence; it is not a telemetry backend.
Runtime guardrails	What must change or stop during a request?	Evaluates at release time; it is not runtime enforcement.
Governance and GRC systems	Which policies, approvals, and accountabilities apply?	Supplies review evidence; it is not a system of record and does not determine compliance.
agent-assure	Did a controlled candidate preserve declared process expectations?	Evaluates, compares when equivalent, packetizes, and returns a CI signal.

It is a particularly strong fit when release review must be local, reproducible, CI-enforceable, and traceable without a required hosted control plane.

How it works

Declare → Capture from a declared source (privacy-filtered) → Evaluate
        → Compare (when equivalent) → Packet → Gate

Declared expectations and canonical run evidence remain distinct. The candidate is evaluated first; equivalent-baseline context is added only after comparison prerequisites pass. The evidence packet then supports a CI signal and human release review.

agent-assure is a local-first Agent Release Assurance Compiler: controls and evidence stay bound to method identity, prerequisites, provenance, assumptions, and limitations.

The assurance model is deliberately bounded:

• Deterministic and reproducible in fixture mode: fixed, versioned fixtures, canonical serialization, schemas, and digest-bound manifests make checks repeatable; that reproducibility does not estimate production prevalence.
• Scoped invariance claims: results cover only declared, observable fields and explicit prerequisites—not hidden reasoning or all production behavior.
• Traceable lineage: expectations connect to RunSets, findings, comparisons, evidence packets, and configured gate state; provenance records participating material.
• Fail-closed: malformed, conflicting, incompatible, ambiguous, or unbound evidence does not silently become a passing review.
• Statistically bounded: live conclusions about probabilistic provider behavior remain tied to a declared statistical protocol, with its data boundary, configuration, window, sampling noise, dependence, and limitations explicit.

Architecture choices and evidence boundaries are documented in architectural decision records (ADRs), including deterministic fixture versus stochastic live semantics.

Challenge the assurance controls

The development-RFC core/v1 mutation catalog runs seven deterministic challenges across evidence linkage, human-review routing, tool boundaries, provenance identity, privacy redaction, duplicate replay, and budget-stop integrity:

agent-assure controls mutate \
  --suite assurance/suite.yaml \
  --runset runs/baseline.json \
  --catalog core/v1 \
  --seed 0 \
  --today 2026-08-02 \
  --full-report \
  --out reports/control-challenge

Every selected operator runs independently against the same immutable source. The output binds the canonical catalog digest, normative expected detector, observed and prohibited substitute findings, exact changed paths, provenance, independence class, seed, and limitations. The campaign itself remains a finite challenge record. controls efficacy can derive an exact catalog-relative detector kill ratio over completed applicable outcomes while preserving inapplicable, invalid, and error counts outside its denominator. That ratio is not a safety score, statistical confidence interval, or universal-coverage claim.

Only deterministic caught and survived campaign outcomes may contribute a control-efficacy verdict. An applicable critical threat with no completed challenge emits CRITICAL_THREAT_UNCOVERED and requires review under the default profile. Required and critical survivors, invalid/error outcomes, and required non-verdict outcomes are always blocking. Evidence-packet ci gate requires control-efficacy evidence by default and uses strict verification when it is present; pass a verifier-owned controls-mutation YAML with --efficacy-policy. --require-efficacy remains as an explicit restatement for existing automation. Strict CI accepts only complete, all-caught, all-applicable-challenged evidence and pins the catalog, selected and required operators, and threat manifest. The conspicuous --allow-missing-efficacy-for-migration escape hatch is only for temporary non-assurance migration and records that weaker profile. Use --allow-advisory-efficacy only for an explicit review flow with evidence present.

For an efficacy-bearing release claim, ci gate --release-profile additionally requires an evidence packet, a verifier-owned --efficacy-policy, present efficacy evidence, and blocking warning/not-evaluated handling. It rejects the advisory and compatibility weakening flags. This is a CI efficacy profile, not publication authorization. make release-publish-check runs that strict profile against the separately staged release efficacy packet and verifier policy before the empirical-readiness and engineering release checks; all are necessary, and none alone authorizes publication.

The packaged offline demonstration exercises both a strong and deliberately weakened assurance control, then verifies that an unrelated failure cannot substitute for the expected detector:

agent-assure demo assure-the-assurance \
  --out .tmp/demo/assure-the-assurance \
  --clean

For a minimal editable workflow, start with agent-assure init controls-mutation, run the read-only doctor controls-mutate preflight, then produce a campaign and control-efficacy-report.json.

Inspect the exact seven-operator catalog · Measure control efficacy · Read "Who assures the assurance?" · Review the evidence contracts

Integrate your agent

agent-assure integrates through declared YAML expectations, versioned run evidence, the documented CLI, and the framework-neutral AgentRunRecord producer contract.

You provide	agent-assure does	You receive
Declared expectations, a candidate RunSet, and an optional equivalent baseline RunSet	Validate, evaluate, compare when equivalence prerequisites pass, packetize, and apply the configured gate	Evaluation and comparison summaries, evidence-packet.json, human-readable reports, and an ordinary CI exit status

The integration contract has three parts:

1. Declare structured process-evidence expectations in YAML.
2. Produce versioned run records from fixtures or privacy-filtered adapters with explicit field origins.
3. Evaluate the candidate, compare equivalent baseline evidence when available, and gate the resulting evidence packet.

A real expectation from the bundled flagship suite:

cases:
  - case_id: shared-source-multi-claim
    fixture_id: shared-source-multi-claim
    expectation:
      expected_recommendation: approve
      required_evidence_refs:
        - ref-shared-clinical-note
      material_claim_ids:
        - claim-eligibility
        - claim-duration

agent-assure does not infer material claims from rationale text. Authors declare the oracle, and run-record producers emit explicit claim-to-evidence links for the material claims they intend to satisfy.

Author expectations · Understand the CLI contract · Review the public API surface · Understand evidence packets

GitHub Actions example using the bundled fixture

Pin both the package and composite action in release workflows. The example uses the latest published tag, v0.6.5; move both pins together only after a newer tag is published. Replace the example suite and variant paths with your own controlled materials.

name: agent-assure
on: [pull_request]

jobs:
  assure:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-python@v5
        with:
          python-version: "3.11"
      - run: python -m pip install agent-assure==0.6.5
      - uses: acblabs/agent-assure/.github/actions/agent-assure@v0.6.5
        with:
          suite: examples/prior_auth_synthetic/suite.yaml
          baseline-variant: examples/prior_auth_synthetic/variants/baseline.yaml
          candidate-variant: examples/prior_auth_synthetic/variants/candidate_evidence_normalization.yaml
          report-mode: full

full produces the complete review artifacts; fail-fast gives shorter blocking feedback. The published v0.6.5 action shown above predates the efficacy-required default and is a fixture smoke example, not evidence of control efficacy. After v0.6.6 or a later version is published, move both pins together. Its composite action fails closed because it does not construct control-efficacy evidence; fixture-only, evaluation-only migration jobs must explicitly set allow-missing-efficacy-for-migration: "true". An assurance workflow must instead construct an efficacy-bearing packet and gate it with a separate verifier-owned policy. The configured gate follows declared expectations and policies, the selected gate profile, and explicit strictness flags. The composite action uploads only the packet, its privacy-filtered assurance evidence graph, manifest, summaries, and CI diagnostics by default. Set upload-full-artifacts: "true" only when the workflow is approved to retain compiled suites, fixture data, and RunSets; the default retention period is 14 days.

Integrations and maturity

Current published release: v0.6.5 on GitHub and PyPI. This checkout is the unreleased 0.6.6 candidate and must not be described or installed as a published release until the empirical publish gate passes.

The CLI, YAML authoring format, persisted versioned JSON artifacts, and AgentRunRecord producer contract are the primary integration surface. Framework adapters, streaming, and live execution remain experimental. The RC label applies only to the primary surface; development-RFC contracts remain non-stable. PyPI's Development Status :: 4 - Beta is the closest standardized classifier to an RC and does not widen that surface.

If you have…	Start with…	Maturity
YAML suites or versioned JSON artifacts	CLI contract	Primary supported surface
A GitHub release workflow	Composite action	Packaged and documented
Deterministic mutation operators and closed catalog campaigns	Core mutation catalog · Evidence-carrying releases	Development RFC
RAG retrieval evidence	RAG provenance demo	Reference implementation
JSONL or multi-agent events	Streaming example	Experimental
LangGraph or Google ADK events	LangGraph · Google ADK	Experimental
Live provider or external-script subjects	Adapter contract	Experimental, time-bound evidence
Preregistered real-model measurement	Real-model study	Untagged development contract; no study result published
Independently controlled CI learning pilot	External pilot evidence	Untagged development contract; no qualifying pilot recorded
OpenTelemetry context or export	OpenTelemetry alignment	Optional alignment only

Want to help with the still-unmet external evidence checkpoint? The short external pilot quickstart targets 10–15 minutes of participant effort in a non-maintainer-controlled fork; CI runtime may be longer. It needs no participant-supplied model key, proprietary data, or repository secret. Interest or a workflow run is not evidence completion: capture, participant friction/finalization and prospective consent, and a byte-bound human review are all required for the pilot bundle.

Experimental streaming semantics

Here, idempotency refers only to idempotent deduplication for stable at-least-once redeliveries. Conflicting duplicates fail closed; deterministic sorting prevents out-of-order arrival jitter from changing the persisted trajectory.

Framework adapters project only privacy-filtered agent_assure metadata into the framework-neutral run-record model. They ignore raw prompts, messages, completions, tool arguments, token chunks, and unredacted summaries.

Governance crosswalks

Packet-resident evidence can be mapped to selected concepts in the NIST AI RMF, OWASP Top 10 for LLM Applications 2025, ISO/IEC 42001, and MITRE ATLAS 2026.06.

These crosswalks are planning and review aids. They do not establish framework conformance, complete coverage, third-party assurance, or endorsement.

Claim boundary

Measured evidence, not a blanket trust claim.

This project is not a compliance attestation.

Generated artifacts make declared inputs, findings, limitations, and gate state traceable and auditable for human review. Whether the release decision is defensible remains a human and organizational judgment. agent-assure does not determine safety or replace domain, legal, regulatory, clinical, security, provider-quality, model-quality, or business-impact review.

agent-assure is	agent-assure is not
Release-review evidence for declared process expectations	A legal or regulatory determination
A deterministic and protocol-bound measurement toolkit	A safety determination
A way to surface evidence, routing, privacy, boundary, provenance, usage, and stream-integrity regressions	A general model-quality benchmark
A local artifact and CI-gate workflow	A production observability backend or enterprise governance system of record
An engineering evidence source for human and governance review	A replacement for organizational accountability

Pattern redaction is a guardrail, not comprehensive DLP or de-identification. Live conclusions remain bounded by the declared protocol, data boundary, provider/model configuration, and execution window. Review the claim boundary, limitations, threat model, privacy model, and security guidance.

Learn more

• Statistical methods: Repeated paired evidence sensitivity · Preregistered real-model study · Live calibration

• Start: Documentation · For AI leaders · For architects · For engineers

• Demos: Assure the assurance · Flagship · RAG provenance · Expense approval

• Integrations: LangGraph · Google ADK · Adapter contract

• Assurance: What this measures · Control efficacy · Evidence packets · Live calibration

• Evidence-carrying releases: Core mutation catalog · Minimal evidence graph · Contracts and campaign guide · Architecture · CLI contract

• Security and governance: Claim boundary · Threat model · Governance crosswalks

• Project: Contributing · Changelog · License

Development from a repository checkout

pip install -e ".[dev]"
git config core.hooksPath .githooks
python scripts/check_docs_alignment.py
ruff check .
mypy src scripts
pytest
python -m build

Citing

This project ships a CITATION.cff. Use GitHub’s Cite this repository control for generated citation formats.