The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources such as logs, codebases, and API responses, which presents an attack surface for indirect prompt injection.
Ingestion points: The skill uses ripgrep, jq, yq, bat, and nushell to read from the local filesystem and external APIs (seen in SKILL.md and references/nushell-structured-data.md).
Boundary markers: There are no instructions or boundary markers to prevent the agent from following instructions embedded within the data it processes.
Capability inventory: The agent is granted capabilities to execute shell commands via fd -x, modify files with sed -i, and perform network operations using xh and Nushell's http commands.
Sanitization: The skill documentation does not provide methods for sanitizing or validating external input before it is used in command pipelines or decision-making processes.
[DATA_EXFILTRATION]: The skill documentation provides examples for locating and extracting sensitive information, which could be combined with its networking tools for exfiltration.
Evidence: references/advanced-regex-and-jq.md includes patterns for extracting environment variables (process.env.(\w+)) and URLs. The skill also facilitates sending data to external endpoints using xh and Nushell's http post command.
[COMMAND_EXECUTION]: The skill facilitates the execution of arbitrary shell commands and multi-file modifications.
Evidence: Examples include using fd -x to execute commands on found files, piping output to xargs, and using sed -i for in-place file refactoring.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for installing various third-party CLI tools.
Evidence: SKILL.md lists installation commands for ripgrep, fd, bat, zoxide, fzf, eza, jq, yq, xh, and nushell via standard package managers like Homebrew, apt, and Cargo. These are well-known developer tools and services.

cli-power-tools