The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches the Bumblebee security scanner CLI from Perplexity AI's official GitHub repository using the Go package manager.
Evidence: go install github.com/perplexityai/bumblebee/cmd/bumblebee@latest in SKILL.md.
[COMMAND_EXECUTION]: Executes shell commands to validate the Go environment, run the inventory scanner across user-specified directories, and manage output files in the workspace.
Evidence: Usage of bumblebee scan with various profiles (baseline, project, deep) and system checks like command -v go.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface by ingesting and summarizing scan output (NDJSON) which could theoretically contain malicious instructions if a compromised package or configuration file had a crafted name/metadata.
Ingestion points: The render_report.py script parses line-delimited JSON generated by the Bumblebee binary.
Boundary markers: Absent; the skill relies on the agent's interpretation of the generated Markdown report.
Capability inventory: Shell command execution, local file system read/write, and Python script execution.
Sanitization: render_report.py includes a safe_user_path function that validates paths to ensure operations stay within the expected workspace and prevent directory traversal.
[DYNAMIC_EXECUTION]: Executes a bundled Python helper script (render_report.py) at runtime to process raw scan data into a human-readable format.
Evidence: python3 "$BUMBLEBEE_SKILL_DIR/scripts/render_report.py" call in SKILL.md.

bumblebee