The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the wahoo-pp-cli tool using npx (from @mvanhorn/printing-press-library) and go install (from github.com/mvanhorn/printing-press-library). These resources are owned by the skill's vendor.
[DATA_EXFILTRATION]: The CLI tool supports a --deliver webhook:<url> flag, allowing users or agents to route command output to arbitrary external HTTP endpoints. Additionally, a feedback command exists which can transmit local data to a remote endpoint if the WAHOO_FEEDBACK_ENDPOINT environment variable is configured.
[COMMAND_EXECUTION]: The skill requires the execution of the wahoo-pp-cli binary to perform its core functions and uses shell commands for initial installation and environment verification.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from an external source (Wahoo Cloud API) and has access to powerful tools.
Ingestion points: Ride data, routes, and user profiles fetched from the Wahoo Cloud API are processed and displayed by the agent.
Boundary markers: The instructions do not specify any delimiters or warnings to the agent to ignore potential instructions embedded within the fitness data.
Capability inventory: The wahoo-pp-cli possesses file system write capabilities (via the backup, profile, and file:<path> delivery features) and network capabilities (via webhooks and feedback reporting).
Sanitization: No sanitization or validation of the content retrieved from the Wahoo API is described, which could allow maliciously crafted fitness data to influence the agent's behavior.

pp-wahoo