strategic-compact
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements context management logic using a local bash script (
suggest-compact.sh). The script tracks tool usage by incrementing a counter in a session-specific temporary file located at/tmp/claude-tool-count-${SESSION_ID}. This is a standard and safe practice for maintaining local state across script invocations. - [SAFE]: The skill configuration uses standard agent hooks (
PreToolUse) to execute the local tracking script. This does not involve executing untrusted or remote commands. - [SAFE]: No network operations, data exfiltration, or credential access were detected. The skill only interacts with a specific temporary file created by the script itself for counting purposes.
- [SAFE]: The instructions and scripts are consistent in their stated purpose of context management and do not contain any malicious patterns such as prompt injection or obfuscation.
Audit Metadata