Do not file public GitHub issues for security reports.

Open a GitHub Security Advisory against the lingzhong/agent-readiness-action repo (Security tab → Report a vulnerability) to start a private disclosure thread. You'll receive an acknowledgment within 72 hours.

This Action takes a URL as user input and calls an external scanner API. The main concerns we design around:

scanner-endpoint is an action input and defaults to the public isitagentready.com API, but can be overridden. A workflow author who overrides this to an internal URL is choosing to do so — same trust level as any other network call from their workflow. We don't additionally restrict the endpoint. Do not accept scanner-endpoint from untrusted pull_request_target inputs.

The scanner response is user-controllable content from a third party. Before emitting ::error:: annotations or writing to $GITHUB_OUTPUT, the script escapes %, \r, and \n per GitHub's workflow-command spec and frames multi-line values in heredoc blocks with a per-invocation delimiter.

The raw response output is truncated at ~900 KB to stay under GitHub's ~1 MB per-output-value limit. A malicious scanner cannot exhaust output storage.

The scanner response is parsed as JSON with jq and specific fields are read by name. No eval, no shell interpolation of response content.

Pre-1.0: only the latest v0.1.x receives security fixes. The floating tag v0.1 tracks the latest patch.