The Agent Skills Directory

[SAFE]: The skill performs read-only analysis of local configuration files (package.json, docker-compose.yml, .env, etc.) and uses the official GitHub CLI (gh) for remote repository discovery. All external operations target the user's own GitHub organization or well-known development paths.
[COMMAND_EXECUTION]: The skill uses Bash for local filesystem discovery and signal scanning. These commands are restricted to the local repository context and common development directories (e.g., ~/git, ~/projects) to identify related services. Path variables are double-quoted to prevent injection, and tool usage is consistent with the stated purpose of ecosystem mapping.
[EXTERNAL_DOWNLOADS]: The skill utilizes the GitHub API via the gh CLI to search for related repositories within a user-specified organization. This is a primary function of the skill and uses official channels for data retrieval. It includes appropriate rate-limiting and error-handling logic.
[PROMPT_INJECTION]: No malicious prompt injection patterns were detected. The instructions focus on the logic for signal detection, confidence scoring, and user confirmation for the discovery process.
[DATA_EXFILTRATION]: While the skill reads environment variable keys and configuration files to identify service names, it does not transmit sensitive values externally. Data usage is limited to generating a local ecosystem map and a batch session file for subsequent analysis steps.

discover