Install
$ agentstack add mcp-clayseal-clayseal-receipts ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
Clay Seal Receipts
Verifiable receipts for AI agent actions.
Clay Seal Receipts wraps an agent or tool call and records what was requested, what policy was checked, which identity or capability context was bound to the decision, and what happened. The result is a receipt that another service, partner, or auditor can verify later without trusting your application logs.
The Python package is named clayseal-receipts. The Python import path remains agentauth.receipts for compatibility.
Install
pip install "clayseal-receipts[server,verifier] @ git+https://github.com/clayseal/clayseal-receipts.git@v0.5.2"
For local development from this repo:
python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
pytest python/tests -q
The identity seam tests need the optional identity package:
pip install -e ".[identity]"
pytest sdk/python/tests -q
Receipts is standalone. It ships the small agentauth.core contract layer it needs, so you do not need a private Clay Seal dependency to use it. Native Clay Seal identity and capability binding are optional extras:
pip install "clayseal-receipts[identity] @ git+https://github.com/clayseal/clayseal-receipts.git@v0.5.2" # Clay Seal identity sessions
pip install "clayseal-receipts[frameworks] @ git+https://github.com/clayseal/clayseal-receipts.git@v0.5.2" # common agent framework adapters
pip install "clayseal-receipts[mcp] @ git+https://github.com/clayseal/clayseal-receipts.git@v0.5.2" # MCP examples and gateway pieces
pip install "clayseal-receipts[opa] @ git+https://github.com/clayseal/clayseal-receipts.git@v0.5.2" # OPA authorization provider
Quickstart
from agentauth.receipts import AgentWrapper, Policy
policy = Policy.from_yaml("policies/fraud_decision.yaml")
wrapper = AgentWrapper(model=my_model, policy=policy, mode="shadow")
result = wrapper.run({"transaction_id": "tx-1", "amount": 50000})
report = result.proof.verify()
assert report.valid
shadow mode records receipts without blocking actions. Move selected workflows to bounded_auto only after you have reviewed receipts from real traffic.
Identity Binding
Receipts can bind to Clay Seal Identity or to claims from providers you already verify, including OIDC, SPIFFE JWT, Auth0, AWS STS, Azure AD, Entra, and GCP.
from agentauth.receipts.identity_providers import get_identity_provider
from agentauth.receipts.integration import wrap_with_identity_session
session = get_identity_provider("oidc").build_session(
verified_claims,
evidence_verified=True,
)
wrapper = wrap_with_identity_session(
model=my_model,
policy=policy,
session=session,
mode="shadow",
)
Provider adapters map already-verified identity claims into receipt authority bindings. They do not replace JWT verification, PoP checks, token revocation, or your normal gateway policy.
Authorization Providers
Receipts can enforce actions through a session authorizer or through a swappable provider. Built-in provider names are agentauth, opa, cedar, openfga, and casbin; teams can also register a provider from a Python callable.
from agentauth.receipts.capability_providers import from_callable
from agentauth.receipts.integration import authorizer_from_provider
provider = from_callable(
lambda action, resource, context: action == "read",
name="read_only",
)
authorize = authorizer_from_provider(provider)
What This Is
Clay Seal Receipts is an audit and verification layer. It helps you prove what an agent was allowed to do and what it actually did.
It is not a full sandbox by itself. For production agent systems, use receipts with short-lived identity, online checks for sensitive actions, scoped capabilities, signed bundles, and retention rules for prompt/tool data.
Docs
- [Developer guide](docs/DEV_GUIDE.md)
- [Deployment guide](docs/deployment.md)
- [Trust model](docs/trust_model.md)
- [Framework integrations](docs/framework_integrations.md)
- [HTTP verifier](docs/http_verifier.md)
- [Privacy and data handling](docs/PRIVACY.md)
- [Security policy](SECURITY.md)
Source & license
This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: clayseal
- Source: clayseal/clayseal-receipts
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.