The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates reading and returning snippets of file contents to the agent via grep:, regex:, and sym: search operators, as well as through a suggested fallback to rg (ripgrep).
Ingestion points: File content matches and text snippets returned by the fsearch and rg tools.
Boundary markers: The skill instructions do not specify any delimiters or warnings to help the agent distinguish between search results and its own instructions.
Capability inventory: The skill utilizes shell command execution for file system searching and content reading.
Sanitization: There is no mention of sanitizing, escaping, or filtering the retrieved file content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill relies on constructing and executing shell commands for fsearch and rg. It passes user-influenced parameters such as search patterns, extensions, and file paths as shell arguments. While the documentation advises quoting these arguments, the dynamic construction of shell commands poses a risk of command injection if the agent or the platform fails to properly sanitize the input.
[EXTERNAL_DOWNLOADS]: The skill references an external URL (https://raw.githubusercontent.com/noahdunnagan/fsearch/refs/heads/main/README.md) to provide the user or agent with installation instructions. Although it points to a common documentation source, the repository is maintained by an individual user rather than a recognized organization, and following remote instructions for software installation carries inherent risk.

fsearch