Add agentcheck to Guardrails & Compliance - #86
Open
uchit wants to merge 1 commit into
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds
agentcheck, a zero-dependency CLI that produces an audit evidence pack for an agent deployment.A note on placement. This section is described as middleware to enforce business logic and safety policies on inputs and outputs, and agentcheck is not middleware — it is an assessment tool that runs against a deployment rather than sitting in the request path. I put it here because "Compliance" is in the section title and nothing else fit better, but if you would rather it went elsewhere, or warranted a separate Assessment & Evidence section, I am happy either way. Say the word and I will move it.
What it does differently. It refuses to produce a score. Most controls for agentic systems can only be attested by a person, and a tool that quietly converts an attestation into a measurement launders an opinion into evidence. So it reports two counts and never sums them:
A command that runs and fails counts as neither. There is a test asserting the output object exposes no aggregate score field, so the one design promise cannot regress quietly.
Why it works as a CI gate. It exits non-zero only when a pack is indefensible — a control claimed met with no evidence recorded, or claimed met while one of its prerequisites is not met (tool authorisation cannot be met if the agent has no identity of its own to authorise). An honest
unknownpasses, because unknown is the correct starting state and a tool that punishes it teaches people to guess.Against the contribution criteria
Disclosure: I am the author.
npm: https://www.npmjs.com/package/@uchit/agentcheck
Docs: https://hellouchit.com/agents/