The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill involves reading project-specific files like package.json and lockfiles, which is a standard surface for indirect prompt injection.
Ingestion points: Project manifests, lockfiles, and configuration files referenced in references/migration-decisions.md.
Boundary markers: Not explicitly defined for the input files.
Capability inventory: Uses bun CLI for installation, testing, and building as described in references/migration-decisions.md and references/runtime-and-tooling.md.
Sanitization: The skill emphasizes security measures such as reviewing trustedDependencies before execution and avoiding hardcoding tokens.
[EXTERNAL_DOWNLOADS]: References official documentation and installation guides from Bun's website.
Evidence: Links to https://bun.com/blog/bun-v1.4.2, https://bun.com/docs/installation, and other official Bun documentation pages in references/migration-decisions.md.

bun-migration