The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingest untrusted data by reading source code, commit messages, and project configuration files to identify "rot." It lack explicit instructions or boundary markers to prevent the AI agent from mistakenly following instructions that might be embedded within the data it is analyzing.
Ingestion points: The skill uses Read, Grep, and various Bash commands (like git log and git grep) to process the contents of any arbitrary repository.
Boundary markers: There are no defined delimiters or instructions to "ignore embedded prompts" when reading file contents.
Capability inventory: The skill is restricted to read-only file operations and specific git/utility shell commands (git, wc, find). It does not have the ability to mutate the filesystem or access the network directly.
Sanitization: No sanitization or filtering is applied to the content read from the repository before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill's instructions recommend installing and executing several third-party static analysis tools from well-known package registries. These include Python packages (vulture, pyflakes, lizard), Node.js packages via npx (ts-prune, knip, unimported), and language-specific tools like cargo udeps or go vet. These are industry-standard utilities for code maintenance and are treated as trusted sources.

fight-repo-rot