The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the tauri-agent-tools CLI to perform various system and application level operations for troubleshooting. Evidence: Commands like tauri-agent-tools diagnose, tauri-agent-tools forensics, and tauri-agent-tools process-tree are used to interact with the operating system, file system, and running processes. The sidecar tap command allows wrapping and running sidecar executables for isolation and validation.
[DATA_EXFILTRATION]: The skill facilitates the collection of logs, system reports, and process information for debugging. Evidence: The bundle and diagnose commands aggregate system data into archives. The skill explicitly addresses security risks by noting that these tools produce a "redacted dir" and stating that "images require review before sharing," showing a design that incorporates data privacy controls.
[INDIRECT_PROMPT_INJECTION]: The skill provides instructions for the agent to ingest and analyze external data sources such as application logs and OS-level log streams. 1. Ingestion points: tauri-agent-tools logs (merges on-disk logs), tauri-agent-tools forensics (reads DiagnosticReports), and tauri-agent-tools os-logs (filters platform log streams). 2. Boundary markers: The skill mentions tool output conventions like NDJSON and markdown summaries which provide structural boundaries for the agent. 3. Capability inventory: The tauri-agent-tools CLI has capabilities for file reading, process listing, and interacting with a local bridge plugin. 4. Sanitization: The skill documentation notes that "bundle redaction failures prevent publication," indicating a built-in validation step for handling external content.

tauri-debug-quickstart