The Agent Skills Directory

[SAFE]: The skill is a utility designed to help AI agents generate a navigation file for LLMs (llms.txt) following the specification at https://llmstxt.org/. The logic is focused on content discovery, planning, and documentation generation within the local repository context.
[DATA_EXPOSURE]: The skill involves scanning the repository's directory structure and reading the contents of READMEs, specifications, and configuration files. This data discovery is confined to local analysis for the purpose of writing a summary file and does not include instructions to access sensitive system files (e.g., .env, .ssh/) or transmit data to external servers.
[REMOTE_CODE_EXECUTION]: No instructions were found that download external scripts, install unverified packages, or execute remote code. The skill relies on analysis tools already available to the agent to examine the repository.
[PROMPT_INJECTION]: The instructions do not contain attempts to override agent safety protocols, bypass constraints, or extract internal system prompts. The "Primary Directive" and "Analysis Phase" are strictly task-oriented.
[COMMAND_EXECUTION]: While the skill suggests using "appropriate tools" to examine the repository structure, it does not provide or require dangerous shell commands, privilege escalation (sudo), or persistent system modifications.

create-llms