legal-proposal-generator
Pass
Audited by Gen Agent Trust Hub on May 13, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of instructional Markdown text and document templates. No executable scripts, binaries, or system-level configuration files are present in the skill package.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of user-provided legal materials and communication records. A malicious actor could embed instructions within these materials to attempt to influence the agent's final report or extract internal prompt logic. The skill lacks explicit sanitization or boundary markers to isolate this untrusted content.\n
- Ingestion points: Case materials, consultation records, evidence files, and chat logs (referenced throughout SKILL.md and template files).\n
- Boundary markers: Absent. The skill relies on standard Markdown headers within templates rather than security-focused delimiters or explicit instructions to ignore embedded commands.\n
- Capability inventory: No capabilities for network access, subprocess execution, or direct file writing were identified across the skill's file set.\n
- Sanitization: None. The skill focuses on extraction and summarization without implementing filtering or escaping of the ingested data.
Audit Metadata