The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill downloads and executes an installation script from Aqua Security's official GitHub repository using a shell pipe. This is used to install the Trivy security scanner if it is not already present on the system.
[COMMAND_EXECUTION]: The skill invokes several system and language-specific tools to perform its duties and apply remediation steps.
Executes trivy for vulnerability and misconfiguration scanning.
Runs local Python scripts (fdk/tools/supply-watch.py) to monitor external supply chain risks.
Uses package managers (go get, pnpm i, brew) to apply security patches and updates.
[DATA_EXFILTRATION]: The skill performs active network operations to test system security and monitor external infrastructure.
Conducts dynamic connectivity tests via curl to verify API authentication and CORS configurations.
Queries DNS records and Certificate Transparency logs (crt.sh) to detect potential typosquatting or brand impersonation.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it processes untrusted data and possesses significant system capabilities.
Ingestion points: Reads project source code, dependency lockfiles (Go, Node, Python), Dockerfiles, DNS records, and Certificate Transparency logs.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore malicious content embedded within the files or external data it analyzes.
Capability inventory: The skill can execute shell commands (trivy, curl, brew, python), modify source code (dependency updates, Dockerfile edits), and perform network requests.
Sanitization: There is no evidence of sanitization or strict schema validation for the data ingested from external scripts or dynamic API tests before it influences subsequent agent actions.

orca-sec-scans