# Adk Rust Eval Observability Guardrails

> Add evaluation, telemetry, and guardrails to ADK-Rust systems as one quality loop. Use when hardening agent behavior before production release.

- Skill: `zavora-ai/adk-rust-eval-observability-guardrails` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add zavora-ai/adk-rust-eval-observability-guardrails`
- Raw SKILL.md: https://api.skillmd.com/api/skills/zavora-ai/adk-rust-eval-observability-guardrails/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: zavora-ai (https://skillmd.com/u/zavora-ai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/zavora-ai/adk-rust-eval-observability-guardrails

---


# ADK Rust Eval Observability Guardrails

## Overview
Treat eval, telemetry, and guardrails as a single production hardening system.

## Workflow
1. Define measurable eval criteria and failure thresholds.
2. Add telemetry spans for model, tool, and callback paths.
3. Add guardrails for PII/content/schema constraints.
4. Run eval suites and compare outcomes to thresholds.

## Guardrails
1. Block promotion when eval criteria regress.
2. Ensure telemetry includes request correlation IDs.
3. Keep guardrail severity and remediation policy explicit.

## References
- Use `references/eval-telemetry-guardrails.md`.