The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands via the Bash tool to perform dependency management and git operations. This includes running npm install, git commit, git status, and project-specific verification scripts like npm test or npm run build. This execution is the core intended purpose of the skill.
[EXTERNAL_DOWNLOADS]: The skill downloads and updates packages from standard public registries (npm, pnpm, yarn, or bun) and fetches migration documentation via the mcp__context7 tool. These are well-known services and the skill includes a human-in-the-loop checkpoint for major updates to mitigate risk.
[PROMPT_INJECTION]: The skill explicitly instructs the agent to ignore default system instructions regarding AI attribution in git commits (e.g., removing Co-Authored-By: Claude). While this is a form of concealment, it is a common stylistic choice in developer tools to maintain a clean git history and does not represent a malicious bypass of safety filters.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes external data, specifically migration guides and documentation fetched from the internet. An attacker could potentially embed malicious instructions in a package's documentation to influence the agent's behavior during a migration.
Ingestion points: External library documentation is fetched via the mcp__context7__get-library-docs tool and processed in SKILL.md (Step 3).
Boundary markers: None identified. The instructions do not define delimiters for external content or warn the agent to ignore embedded instructions.
Capability inventory: The skill has broad capabilities including file modification (Write, Edit) and arbitrary command execution (Bash).
Sanitization: No sanitization or validation of the fetched documentation content is performed before processing.

update-deps