load-forecast

Pass

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. Ingestion: The skill processes local codebases and user arguments ($ARGUMENTS). Boundary markers: No delimiters or instructions to ignore embedded directives are present in the prompt logic. Capability inventory: The skill is designed for the CLAUDE_CODE platform, granting it shell access and file read capabilities. Sanitization: No sanitization or validation of ingested content is performed before processing.
  • [DATA_EXFILTRATION]: Local Telemetry and Persistence. The 'SELF-EVOLUTION TELEMETRY' section instructs the agent to write execution metadata (outcome, iterations, bottlenecks) to '~/.claude/projects/skill-telemetry.md'. This constitutes a mechanism for cross-session tracking and persistence within the user's home directory.
  • [COMMAND_EXECUTION]: Discovery Attack Surface. The skill requires the agent to auto-detect system architecture by scanning various configuration files (requirements.txt, pom.xml, docker-compose.yml). In a shell-enabled environment, performing discovery on untrusted file structures can be exploited if the agent misinterprets malicious file content or filenames as shell commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 23, 2026, 10:56 AM
Security Audit — agent-trust-hub — load-forecast