The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill performs an audit of repository-controlled files to guide migration actions, which exposes it to indirect prompt injection from the content of the audited code.
Ingestion points: SKILL.md and references/adoption-checklist.md instruct the agent to analyze package.json, lockfiles, imports, and renderer props within the host repository.
Boundary markers: The instructions do not specify boundaries or directives to ignore instructions that might be embedded within the analyzed repository data.
Capability inventory: The skill allows modification of project dependencies and environment state using pnpm commands.
Sanitization: There is no evidence of sanitization or validation performed on the ingested repository content before it is used to determine migration steps.
[COMMAND_EXECUTION]: The skill directs the execution of shell commands to manage project dependencies during the migration process.
Evidence: references/vue-1x-to-2x.md specifies commands such as pnpm remove stream-monaco stream-markdown and pnpm add markstream-vue@next.
[EXTERNAL_DOWNLOADS]: The migration process involves downloading and installing external packages from a package registry.
Evidence: The skill guides the installation of various markstream-* adapter packages and peer dependencies such as octane and svelte.

markstream-migration