The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could contain instructions designed to influence the model's behavior.
Ingestion points: The tool ingests data from arbitrary URLs using the -u flag and YouTube transcripts using the -y flag.
Boundary markers: While custom patterns use headers like '# INPUT:', there are no explicit delimiters or 'ignore' instructions for raw web content.
Capability inventory: The tool interacts with LLM providers (OpenAI, Anthropic, Gemini) and local Ollama servers.
Sanitization: No sanitization or filtering of the ingested content is mentioned.
[COMMAND_EXECUTION]: The instructions require the execution of various shell commands for setup and use.
Evidence: Commands include 'go install', 'brew install', 'winget install', 'curl', 'tar', and the 'fabric' binary itself.
[REMOTE_CODE_EXECUTION]: The installation process involves downloading a pre-compiled binary from a remote repository and executing it.
Evidence: Fetches a release archive from 'github.com/danielmiessler/Fabric' and executes the extracted binary.
[EXTERNAL_DOWNLOADS]: The tool downloads pattern libraries and updates from its official source.
Evidence: The 'fabric -U' command refreshes built-in patterns from the remote repository.

fabric