The Agent Skills Directory

[DATA_EXFILTRATION]: The coffee-goat-pp-cli tool documented in the skill supports a --deliver flag that can route output to local files (file:<path>) or remote servers (webhook:<url>).
Evidence: The 'Output Delivery' section explicitly defines these sinks, allowing the agent to POST command results (including private cellar logs and palate profiles) to arbitrary URLs.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Data enters the context via creator-review (YouTube transcripts), transcript-search (Hoffmann/Hedrick video text), and search (external roaster catalogs).
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the processed coffee data.
Capability inventory: The agent has access to the Bash tool and the CLI's own file-writing and network-POSTing capabilities.
Sanitization: No sanitization or validation of the external transcript or catalog content is documented.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of a CLI tool from external registries.
Evidence: The installation instructions use npx -y @mvanhorn/printing-press-library and go install github.com/mvanhorn/.... These resources are owned by the skill's author (mvanhorn).
[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands to install, verify, and run the coffee-goat utility.
Evidence: Multiple bash blocks contain commands for tool installation and feature execution (e.g., coffee-goat god-cup, coffee-goat search).

pp-coffee-goat