The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/agent.py script contains a check_dependency_confusion function that uses subprocess.run to execute npm and pip commands. While this function is intended for legitimate dependency validation against public registries and is not currently called in the script's main workflow, it represents a command execution surface that handles package names from the environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data sources, such as build logs and software dependency manifests, which creates a vulnerability surface for indirect prompt injection.
Ingestion points: scripts/agent.py (parsing package-lock.json, requirements.txt, and build logs) and scripts/process.py (parsing JSON/CSV logs).
Boundary markers: Absent. The scripts do not use specific delimiters or instructions to prevent the agent from following malicious instructions that might be embedded in the analyzed logs.
Capability inventory: The skill includes capabilities for file system read/write access and shell command execution via subprocess.run in scripts/agent.py.
Sanitization: The skill utilizes regex-based matching for detection but does not perform sanitization or escaping of the ingested content before incorporating it into the final Markdown hunt reports.

hunting-for-supply-chain-compromise