The Agent Skills Directory

[SAFE]: The skill follows security best practices for developer automation tools. It avoids hardcoding sensitive credentials, instead relying on environment variables (e.g., LOVSTUDIO_REVALIDATE_SECRET) and established CLI authentication (gh auth). The provided Python scripts utilize yaml.safe_load() to mitigate risks associated with parsing untrusted YAML data, and the documentation explicitly instructs the user to keep platform metadata and secrets outside of the canonical source code.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands to automate the release process.
Evidence: The workflow utilizes git, gh, python3, and curl to initialize repositories, tag releases, build packages, and trigger site revalidation. These commands are legitimate for the skill's stated purpose of being a publisher and are performed using established, well-known software tools.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from other local skill files, which creates an attack surface for indirect prompt injection, although it is mitigated through validation.
Ingestion points: The publishing process involves reading SKILL.md and kit.yaml from local directories to extract metadata like names, descriptions, and versions.
Boundary markers: The skill workflow includes a mandatory validation step (scripts/validate_skill.py) that must pass before any publishing actions are taken.
Capability inventory: The skill possesses the ability to execute shell commands, perform network requests to specific endpoints, and write to the local file system (e.g., creating ZIP archives).
Sanitization: The scripts/validate_skill.py script enforces strict naming (kebab-case) and versioning (SemVer) rules via regex, and utilizes yaml.safe_load to prevent code execution during file ingestion.

lov-skill-publisher