ai-short-drama-master

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill implements an 'Agentic Protocol' that directs the agent to fetch and process external data, creating a surface for indirect prompt injection.
  • Ingestion points: Step 2 of the Agentic Protocol in SKILL.md instructs the agent to use WebSearch, WebFetch, and agent-reach to gather industry information.
  • Boundary markers: Absent; there are no instructions to use delimiters or ignore embedded instructions within the fetched data.
  • Capability inventory: The skill includes bash scripts (cli/protocol/agentic.sh, cli/decision/evidence.sh) that can write files to the local system.
  • Sanitization: No validation or sanitization rules are specified for the data retrieved from external sources.
  • [COMMAND_EXECUTION]: The skill includes a CLI component with several bash scripts (cli/protocol/agentic.sh, cli/decision/evidence.sh, cli/decision/topic-2.sh) and a shared library (cli/lib/common.sh). These scripts facilitate interactive workflows, read multi-line user input, and use the printf command to write markdown reports to the local file system. This is intended functionality for the 'Master OS' pattern but involves direct shell interaction.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 03:01 PM
Security Audit — agent-trust-hub — ai-short-drama-master