kata-pause-work
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses local system tools like
grep,find, andgitto manage project state within the.planningdirectory. It also invokes a local Node.js utility script (scripts/kata-lib.cjs) for reading and writing configuration settings. All execution is restricted to the local environment and serves the skill's primary function of task management.- [INDIRECT_PROMPT_INJECTION]: The skill reads from local files including.planning/STATE.mdand.planning/config.jsonto determine the current work phase. - Ingestion points:
.planning/STATE.mdand project configuration files. - Boundary markers: None explicitly defined in instructions.
- Capability inventory: File writing, Git commits, and shell command execution for discovery.
- Sanitization: Extraction of the current phase uses a strict numeric regular expression (
grep -oP '[0-9]+'), which prevents non-numeric injection from influencing path construction.- [SAFE]: No evidence of credential theft, data exfiltration, or unauthorized remote code execution was found. The skill follows standard practices for project-level state persistence.
Audit Metadata