The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The installation process requires cloning an external GitHub repository (github.com/sparkfinderoven/r01-hesreallyhim-awesome-claude-code-security.git) and executing binaries located in its bin/ directory. This allows for the execution of unverified code at runtime.
[EXTERNAL_DOWNLOADS]: The skill relies on external software components and binaries fetched from a repository that is not associated with recognized trusted vendors or organizations.
[DATA_EXFILTRATION]: The skill contains commands and examples for sending detailed security audit results, vulnerability findings, and compliance data to external webhooks (e.g., Slack) and APIs (e.g., Jira). This creates a path for sensitive internal architectural and security data to be transmitted outside the local environment.
[COMMAND_EXECUTION]: Multiple commands utilize shell features such as pipes, environment variable expansion, and command substitution. For instance, integration examples pipe JSON scan output directly into curl requests to external servers.
[PROMPT_INJECTION]: The skill has a significant indirect prompt injection surface because its primary functions involve reading and analyzing untrusted external data, such as project source code, architecture documentation, and dependency manifests. Malicious instructions hidden in these files could manipulate the agent's audit findings or behavior.
Ingestion points: Local filesystem (source code files, architecture .md and .yaml files, and data inventory JSONs).
Boundary markers: The skill does not provide the agent with specific delimiters or instructions to ignore potential commands embedded in the files it scans.
Capability inventory: The skill possesses file read access, the ability to execute local binaries, and network access via curl.
Sanitization: There are no mechanisms described to sanitize or validate the content of the files before they are processed by the agent.

sparkfinderoven-security-compliance-suite