internal-comms

Pass

Audited by Gen Agent Trust Hub on Apr 15, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The instructions in 'examples/3p-updates.md', 'examples/company-newsletter.md', and 'examples/faq-answers.md' direct the agent to summarize content from external sources such as Slack, Email, and Google Drive. This creates a surface for indirect prompt injection where malicious instructions embedded in those messages could attempt to influence the agent's output.
  • Ingestion points: Slack messages, emails, and Google Drive documents identified in the 'Tools' sections of the example files.
  • Boundary markers: The skill prompts do not use delimiters or provide instructions to ignore potentially conflicting directives within the retrieved data.
  • Capability inventory: No scripts, subprocess calls, or network operations are defined in the skill itself; it relies on the agent's pre-configured tool access.
  • Sanitization: There is no logic provided to filter or escape the gathered content before processing it into summaries.
  • [NO_CODE]: The skill consists entirely of Markdown files and a license text. No executable scripts, binaries, or configuration files that trigger code execution were detected in the package.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 15, 2026, 06:10 PM
Security Audit — agent-trust-hub — internal-comms