Back to Gap Map Product / UX / Agent identity & authorization

Bolyra

ZKProva
open source / Overall score: 2.2

Bolyra is a gateway and verifier that checks each AI-agent tool call before it reaches an MCP server. It verifies the agent's credential, applies per-tool policy, rejects replays and signs a receipt for every allow or deny. An optional zero-knowledge tier lets agents prove credentials and one-way narrowed delegations without revealing them. ZKProva Inc. builds it.

Openness

5 high confidence
5.0
license
Apache-2.0 (gateway, SDKs, circuits, contracts and integrations)
source
public
core features withheld
no — the 3,873-path tree at f4475fa carries the root Apache-2.0 LICENSE and 16 package LICENSE files, all Apache-2.0, with no ee/, enterprise/, commercial/ or proprietary/ path; the vendor sells a paid acceptance exercise, not an edition
commercial
acceptance exercise and managed-revocation pilot (services around the self-hosted gateway)

Bolyra's gateway, SDKs, circuits, contracts and integrations are all Apache-2.0 code in one public repository. ZKProva says the gateway is self-hosted inside a customer's own network, and what it sells is a paid acceptance exercise and pilots, services rather than a withheld edition.

Adoption

1 medium confidence
1.0

Adoption is read from npm downloads of the Bolyra TypeScript SDK and the MCP gateway. The Python package, a thin bridge to the TypeScript code, is not counted, and downloads cannot show how many gateways guard a live server.

Capability

4 medium confidence
4.0

Bolyra verifies the agent's own credential, applies per-tool policy to every call, narrows delegated permissions one way and signs a hash-chained receipt for each decision. Its gateway does not hold a call while a person reviews it.

Verified 2026-11-20, except some scores, still awaiting re-confirmation