The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's automatic update workflow is designed to ingest data from GitHub release payloads, which is explicitly described as "untrusted text".
Ingestion points: GitHub release payload / event text during automated triggers.
Boundary markers: None specified for the untrusted release content.
Capability inventory: The skill has significant local capabilities including executing pnpm (package installation), git (file mutation/commits), and node (script execution).
Sanitization: No sanitization or validation of the release payload text is mentioned before it is processed by the agent.
[COMMAND_EXECUTION]: The skill executes several CLI tools to manage the update process.
Evidence: Uses pnpm for package resolution and installation, git for repository state management and diffing, and gh for interacting with GitHub releases.
Evidence: Runs a local script node scripts/check-plugin-packages.mjs to verify configuration during the update process.
[EXTERNAL_DOWNLOADS]: The skill fetches external code and metadata as part of its primary function.
Evidence: Clones the getsentry/junior repository to perform configuration diffing against upstream examples.
Evidence: Downloads and installs @sentry/junior and related plugin packages from the npm registry via pnpm add.

self-update