ambit-cli

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt includes examples and flags that embed secrets directly in commands (e.g., --api-key <key> and secrets set my-app.lab API_KEY=abc123), which instructs the agent to handle and output secret values verbatim, creating an exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md explicitly describes a "template mode" that fetches arbitrary GitHub repositories via GitHub's tarball API ("--template owner/repo"), extracts and pre-flight scans their fly.toml, and then deploys them—meaning untrusted, user-authored third-party content is ingested and can directly change deployment actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill invokes npx @cardelli/ambit (which fetches and executes code from the npm registry at runtime) and explicitly fetches GitHub templates such as ToxicPine/ambit-templates/chromatic via the GitHub tarball API (e.g. https://github.com/ToxicPine/ambit-templates), so remote content is fetched during runtime and executed as a required dependency.