The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external websites and local codebase files during its research phase to gather context for prompt improvement.
Ingestion points: The workflow detailed in references/research-strategies.md uses WebSearch, WebFetch, Glob, Grep, and Read tools to access potentially untrusted content from the internet and the local filesystem.
Boundary markers: The instructions in SKILL.md and the associated reference files do not prescribe the use of delimiters or specific directives (such as "ignore embedded instructions") when processing the gathered research content.
Capability inventory: The agent maintains access to powerful capabilities including command execution via Bash (used for git and npm test in examples) and file-writing via TodoWrite as outlined in SKILL.md and references/examples.md.
Sanitization: No sanitization or validation of the data retrieved from external sources is specified before it is used to ground questions or execute the final transformed request.

prompt-improver