The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the git log command to extract commit and PR data. This is a benign use of system commands for the skill's primary function.
[SAFE]: The skill reads local project files (.acumen/features.md, .acumen.md) to ground its analysis in business context. This file access is legitimate and restricted to the project directory.
[SAFE]: The skill processes untrusted data from commit messages and pull requests (Ingestion point: git log output). It lacks explicit boundary markers to separate this data from instructions. The capability inventory is limited to local shell commands and file reading (Capability inventory: git log), with no network or write access. No sanitization of the git history is performed (Sanitization: absent).

changelog