The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill is designed for observability and strictly adheres to the principle of least privilege.
[DATA_EXPOSURE]: The skill accesses project-specific state files (pdca-status.json and token-ledger.ndjson) to render the dashboard. This file access is read-only and restricted to the .bkit directory, which is consistent with its stated purpose of monitoring internal workflow state.
[COMMAND_EXECUTION]: The skill explicitly disables subprocess spawning and does not request tools capable of executing shell commands. It relies on the platform's native /loop mechanism for periodic updates rather than custom scripts.
[DATA_EXFILTRATION]: The skill does not request network access tools and operates entirely within the local environment. It includes self-imposed bounds on data processing (e.g., limiting log tailing to 200 lines) to prevent resource exhaustion.

pdca-watch