Skip to content
@Signetry

Signetry

Seal every agent's PR with proof — a change-control plane for coding agents. Earned authority, sealed in a signed receipt.

Signetry

Signetry

Seal every agent's PR with proof — earned authority in a signed receipt.


Signetry is a change-control plane for coding agents. Any agent (Claude Code, Codex, Cursor, Copilot, Devin — or a human) may propose a change; only Signetry decides how much authority that change earned, and seals the verdict in an Ed25519-signed receipt.

For every change it runs one deterministic pipeline:

executable contract  →  untrusted-text quarantine  →  required checks  →
independent verifier  →  earned authority (0/1/2)  →  Ed25519-signed receipt

A coding agent cannot approve its own authority to make a change. The patch-writer is never the patch-approver. auto_merge is always false — a human merges.

Start here

Repo What it is
signetry The overview / front door — architecture, integrations, compatibility
core The governance kernel (signetry-core, the signetry CLI)
action Signetry Admission — govern every PR on the GitHub Marketplace
reviewer Advisory PR reviewer with a deterministic merge-safety gate
eval Public adversarial benchmark (attack-success-rate + utility)
plugins · claude-code · codex · cursor · precommit Editor / agent / hook integrations

Install

# source-available (All Rights Reserved); installed from source, not PyPI
pip install "signetry-core @ git+https://github.com/Signetry/core@v0.6.0"

Contributing

Signetry is source-available (All Rights Reserved, © 2026 Binay Dalai) — not open source. Contributions are welcome under a Contributor License Agreement: you're credited, but gain no ownership or right to use/sell it. See any repo's CONTRIBUTING.md and CLA.md.

Pinned Loading

  1. core core Public

    Agent-agnostic change-control plane for coding agents. Govern Claude Code, Codex, Cursor, or any agent behind one admission pipeline — prove every change with a signed receipt.

    Python 1 3

  2. signetry signetry Public

    The umbrella overview for the Umbra platform — architecture, integration catalog, compatibility matrix, and release train. Governance code lives in umbra-core.

  3. eval eval Public

    The public adversarial evaluation suite for Umbra: measures attack success rate (ASR) and utility-under-defense for coding-agent prompt injection, skill/MCP poisoning, and memory-injection threats,…

    Python

Repositories

Showing 10 of 15 repositories

Top languages

Loading…

Most used topics

Loading…