Skip to content

mizcausevic-dev/agent-card-fleet-summary-action

Use this GitHub action with your project
Add this Action to an existing workflow or create a new one
View on Marketplace

Repository files navigation

agent-card-fleet-summary-action

CI License: AGPL-3.0-or-later

GitHub Action that walks a directory of A2A AgentCard documents, surfaces the governance gaps that hurt an audit (autonomous-without-IRU, destructive-tool-on-non-autonomous, persistent-memory-without-refusal-taxonomy, etc.), posts a Markdown summary as a PR comment, and fails the build when any high-severity finding is present.

Wraps agent-card-fleet-summary — same finding logic, vendored into the action for self-contained execution.

Part of the Kinetic Gain Suite. Sibling of llm-cost-rollup-action and k8s-pre-merge-action.


Usage

name: A2A governance
on:
  pull_request:
    paths: ["agents/**"]

jobs:
  fleet-summary:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: mizcausevic-dev/agent-card-fleet-summary-action@v0.1-shipped
        with:
          cards-dir: agents/
          fail-on-high: true   # default

Inputs

input required default description
cards-dir Directory containing *.json AgentCard documents.
comment-on-pr auto Post the Markdown summary as a PR comment. auto posts only on pull_request events; true / false force the behavior.
fail-on-high true Fail the run when any high-severity finding is present.
github-token ${{ github.token }} Token used to post the PR comment.

Outputs

output description
total-cards Number of AgentCards analyzed.
high-findings Count of high-severity findings.
autonomous-count Number of autonomous agents in the fleet.
destructive-tools Total destructive tools declared across the fleet.

What it flags

Code Severity Rule
autonomous-without-incident-response-uri 🔴 Autonomous agent missing safety_posture.incident_response_uri.
no-evaluations-on-autonomous 🔴 Autonomous agent has no evaluations[].
destructive-tool-on-non-autonomous 🟠 Non-autonomous agent declares destructive tools.
persistent-memory-without-refusal-taxonomy 🟠 Persistent-memory agent has no refusal taxonomy.
empty-refusal-taxonomy 🟡 No refusal categories declared.
no-evaluations 🟡 No evaluations on a non-autonomous agent.
missing-homepage ℹ️ agent.homepage is not set.

Composes with

License

AGPL-3.0-or-later

About

GitHub Action wrapping agent-card-fleet-summary. Walks a dir of AgentCards, surfaces governance gaps (autonomous-without-IRU, destructive-on-non-autonomous, etc.), posts Markdown PR comment, fails on high findings. Sibling of llm-cost-rollup-action.

Topics

Resources

License

Code of conduct

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Packages

 
 
 

Contributors