The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple system commands including git, jq, grep, and awk. It specifically attempts to execute a project-local validation script at .claude/skills/skill-linter/scripts/validate-skill.sh to check plugin integrity. It also automates the staging, committing, and pushing of code to the origin master branch.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources within the repository that could be influenced by an attacker. 1. Ingestion points: Processes content from plugin.json, marketplace.json, git status output, git diff output, and plugin README.md files. 2. Boundary markers: No explicit delimiters or instructions are provided to the model to ignore potential instructions embedded within these files. 3. Capability inventory: The skill has the ability to Edit files, git add, git commit, and git push changes, and execute local shell scripts. 4. Sanitization: Uses jq with arguments for structured data processing, but other text-based data from git and readmes is processed without explicit sanitization or escaping.

release