pudu-task-telemetry

Topics in the source

• Pudu Task Telemetry
• 1. Diagnose
• 2. Select a model
• 3. Execute a bounded subtask
• 4. Verify and close
• 5. Report honestly
• Sources

Use Pudu AI to inspect hardware and benchmark evidence, execute a bounded text subtask through local Ollama, and report measurements with their provenance. This skill captures its own local calls; it does not observe all activity or change the model of the host assistant. Locate this skill's scripts/pudu-task.mjs relative to this file. Examples assume project installation under .claude/skills/pudu-task-telemetry/. Run from the project root, or supply --repo explicitly. Requires Node.js >=20, Pudu AI on PATH, and a running Ollama server. Diagnose missing dependencies without installing packages, downloading models, or changing global settings. Read setup.md for configuration and the separate server-side local-only prerequisite. A loopback URL alone does not prove that the server cannot forward a request to cloud inference. Prefer installed models that fit the task's context and hardware. A Pudu hardware score is not a quality score. Without a comparable verified suite, recommendations return needsselection; select a model explicitly for a pilot. Read model-selection.md before using --model auto or interpreting comparisons. Do not invent model IDs or claim a universal winner. Prepare

This is an excerpt. Refer to the complete source and its current requirements.