The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill suggests installing zizmor via cargo install zizmor. While a legitimate security tool, this involves downloading and installing third-party software at runtime.
[COMMAND_EXECUTION]: The skill uses python3 -c to execute a script that generates typosquatting candidates based on package names. It also relies heavily on shell scripts for reconnaissance.
[CREDENTIALS_UNSAFE]: The skill actively searches for sensitive configuration files and credentials, such as .npmrc, pip.conf, .env, and _authToken, using recursive grep commands.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface as it processes untrusted data from external sources, including public GitHub repositories, JavaScript bundles, and Docker registries. It uses gh api and curl to fetch content that is then processed by the agent.
[DYNAMIC_EXECUTION]: The skill includes shell commands that decode GitHub API content using base64 -d and process it using jq and grep.

supply-chain-attack-recon