The Agent Skills Directory

[SAFE]: The skill is a well-structured engineering framework that focuses on development best practices such as TDD, code review, and repository auditing. It encourages the use of isolated worktrees and structured documentation for engineering orchestration.
[COMMAND_EXECUTION]: Python and shell scripts (repo_scan.py, harness_audit.py, find-polluter.sh) execute standard developer tools like git, ripgrep, and npm for repository orientation, auditing, and testing. These operations are foundational to the skill's software engineering purpose and do not involve untrusted input or risky execution patterns.
[EXTERNAL_DOWNLOADS]: The skill includes boilerplate for GitHub Actions and GitLab CI that references official, well-known, and trusted actions (e.g., from the actions organization). These are standard configurations for setting up automated testing and build environments.
[DATA_EXFILTRATION]: The skill implements a local learning and memory system (learning/, memory/ directories) to persist knowledge across sessions. It captures tool outputs for error logging and self-improvement, but all data remains within the project's local file system without evidence of network exfiltration.

agentic-development