The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various subcommands of the langwatch CLI (such as trace search, scenario list, analytics query, prompt init, and monitor create) and git commands (git log) to perform a comprehensive audit and management of the agent's development environment. These commands are essential to the skill's intended functionality as an engineering consultant.
[INDIRECT_PROMPT_INJECTION]: The skill requires the agent to "read EVERYTHING" in the codebase, including all source files, functions, prompts, and git commit history. This broad data ingestion creates a vulnerability to indirect prompt injection if the analyzed files contain malicious instructions disguised as code comments, documentation, or commit messages.
Ingestion points: All local codebase files, git commit logs, and tool outputs from the langwatch CLI.
Boundary markers: None identified; the skill does not instruct the agent to use specific delimiters or explicitly ignore instructions within the ingested codebase content.
Capability inventory: File system read access, git command execution, and LangWatch CLI command execution (including operations that modify state, such as monitor create and prompt init).
Sanitization: The skill does not specify any sanitization, filtering, or validation of the codebase content before the agent processes and acts upon the information.

agent-best-practices