The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it is designed to ingest and analyze untrusted external codebases using tools like Read, Glob, and Grep. Evidence Chain:
Ingestion points: The skill instructions direct the agent to analyze project files to detect breaking changes and apply codemods.
Boundary markers: No explicit boundary markers or instruction-override warnings are provided when processing external content.
Capability inventory: The skill uses high-privilege tools including Bash, Write, and Edit to perform code transformations and execute ecosystem commands.
Sanitization: The skill recommends manual review of automated codemod outputs in its 'Common Gotchas' section, which serves as a human review checkpoint.
[EXTERNAL_DOWNLOADS]: The scripts/check-migrate-facts.py utility fetches version metadata from well-known registries including registry.npmjs.org and endoflife.date. These operations are restricted to data retrieval for version validation and do not involve remote code execution. Additionally, the skill suggests the use of standard ecosystem tools such as npx and pip to fetch and run official codemods.
[COMMAND_EXECUTION]: The skill utilizes a wide array of standard developer CLI tools for auditing and refactoring, including npm-audit, pip-audit, cargo-audit, and various framework-specific codemod runners. These are documented as part of a legitimate developer workflow for software maintenance.

migrate-ops