context-budget
Pass
Audited by Gen Agent Trust Hub on Apr 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute file system operations, specifically reading and globbing files, to access local configuration and project-level data for token estimation. Access includes system-level files like ~/.claude/CLAUDE.md, settings.json, and hooks.json.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection during its analysis phase. 1. Ingestion points: The agent reads content from untrusted project files (CLAUDE.md, rules/*.md) and internal configuration files (SKILL.md). 2. Boundary markers: Absent; there are no instructions to delimit or ignore embedded commands within the read data. 3. Capability inventory: File reading and globbing across system and project paths in SKILL.md. 4. Sanitization: Absent; the agent is directed to count characters of raw file content without filtering for malicious patterns.
Audit Metadata