maintenance-scheduling
Warn
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONPERSISTENCE
Full Analysis
- [DATA_EXPOSURE_EXFILTRATION]: The skill attempts to access the agent's internal configuration and state directory at
~/.claude/projects/to locate project memory files. - [PERSISTENCE]: The 'Self-Evolution Telemetry' section instructs the agent to maintain state across sessions by appending execution metadata to
skill-telemetry.mdwithin the internal project memory directory. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process a wide variety of external, untrusted data sources including CMMS configurations, work order records, and IoT/BMS sensor feeds, which could contain malicious instructions.
- Ingestion points: CMMS platforms (Maximo, SAP PM, etc.), asset registries, IoT sensor feeds, and work order data.
- Boundary markers: None identified; instructions do not include delimiters or warnings to ignore embedded content.
- Capability inventory: The skill has the capability to write to the file system (
docs/maintenance-scheduling-analysis.md) and modify internal telemetry logs (~/.claude/projects/.../skill-telemetry.md). - Sanitization: No evidence of data validation, escaping, or instruction filtering was found.
Audit Metadata