cost-session
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill reads local session history files located in
~/.claude/projects/to extract usage statistics. This is the intended purpose of the tool and does not involve unauthorized access to system credentials or network exfiltration. - [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it ingests session log data. 1. Ingestion points:
~/.claude/projects/session logs (SKILL.md). 2. Boundary markers: None provided. 3. Capability inventory: Bash (SKILL.md frontmatter). 4. Sanitization: None described in the provided logic. However, the risk is considered safe as the skill's logic is focused on costing metrics rather than executing message content.
Audit Metadata