The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands to manage project state and assets, including git status, npm run build, npm test, and composer run test. These commands trigger scripts defined in the project's package.json and composer.json files, which are external to the skill itself.
[EXTERNAL_DOWNLOADS]: The procedure allows for the installation of missing dependencies via npm or composer when lockfiles are detected. This involves fetching packages from public registries.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from the repository being edited (e.g., version headers in PHP files, existing changelogs, and package configurations) and incorporates user-provided release notes into files.
Ingestion points: Project files like package.json, composer.json, readme.txt, and PHP headers are read in Step 0 and Step 2 of SKILL.md.
Boundary markers: None are defined to separate untrusted file content from the agent's instructions.
Capability inventory: The skill can perform file writes, run shell scripts (npm/composer), and manage dependencies as described in Steps 2, 4, and 5 of SKILL.md.
Sanitization: No explicit sanitization or validation of the content read from files is performed before it is used to determine subsequent actions.

wp-bump