The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill contains a command that downloads a shell script from a remote URL and pipes it directly into the bash interpreter for immediate execution.
Evidence: In references/install-and-provider-setup.md, the installation step curl -fsSL https://raw.githubusercontent.com/danielmiessler/fabric/main/scripts/installer/install.sh | bash executes unverified remote code in the local shell environment.
[COMMAND_EXECUTION]: The skill relies on the Bash tool to perform installation and run command-line operations for the Fabric tool.
Evidence: Multiple files including SKILL.md and references/pattern-workflow-recipes.md provide shell commands like cat transcript.txt | fabric -p summarize for the agent to execute.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to retrieve software and configuration from external repositories that are not managed by trusted organizations.
Evidence: References to github.com/danielmiessler/fabric for installation scripts and API documentation.
[PROMPT_INJECTION]: The skill processes potentially untrusted external content, creating a surface for indirect prompt injection.
Ingestion points: SKILL.md specifies that the tool processes stdin, files, transcripts, notes, logs, and web text.
Boundary markers: Absent; there are no instructions provided to the agent to distinguish between data and instructions within the processed text.
Capability inventory: The skill has access to the Bash, Read, and Write tools.
Sanitization: Absent; the instructions do not include any validation or escaping mechanisms for the input data before it is processed by the AI transforms.

fabric