The Agent Skills Directory

[METADATA_POISONING]: The skill exhibits inconsistent authorship metadata. The YAML frontmatter in SKILL.md lists the author as mahipal, whereas the LICENSE file attributes copyright to mukul975.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from industrial network traffic summaries and process historian databases, which could be exploited to influence agent behavior.
Ingestion points: External data enters through analyze_modbus_traffic (network traffic summaries) and query_historian_anomalies (historian API responses) in scripts/agent.py.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the data processing logic.
Capability inventory: The skill possesses capabilities for network socket communication, Modbus protocol interaction via pymodbus, and HTTP requests via requests.
Sanitization: While the skill applies statistical Z-score filtering to sensor data, it does not sanitize inputs for prompt injection or malicious patterns.
[DATA_EXFILTRATION]: The script scripts/agent.py performs network operations including Modbus device scanning and historian API queries. The query_historian_anomalies function transmits a Bearer token to a user-provided URL, which could lead to credential exposure if directed to a malicious endpoint.
[EXTERNAL_DOWNLOADS]: The skill references official documentation and security guidelines from trusted sources including NIST, CISA, and the Pymodbus documentation site.

detecting-anomalies-in-industrial-control-systems