Azure Monitor Query for Python

The skill prefers DefaultAzureCredential, discourages connection strings and API keys, and requires deterministic cleanup for clients and async credentials. Its examples are read-oriented for logs and metrics, with no destructive defaults or credential theft. Points are deducted because least-privilege RBAC, sensitive-log handling, data-flow disclosure, user confirmation, isolation, rollback, and dependency security policy are not specified; versions are also unpinned.

The evidence covers sync, async, batch queries, partial-result handling, and lifecycle cleanup, so the main path is plausible. Points are deducted because static review cannot verify execution; examples omit necessary imports such as os, require_envvar=True may conflict with the claim that local DefaultAzureCredential works as-is, and handling for errors, timeouts, throttling, and invalid input is limited.

The skill clearly targets Python Azure Monitor, Log Analytics queries, and Azure metrics, with explicit trigger terms, clients, and sync/async scenarios. Points are deducted because non-fit boundaries, permission prerequisites, SDK compatibility ranges, Chinese-language guidance, and mainland-China network fit are not documented; service access may depend on region, network, and Azure configuration.

The evidence includes front matter, MIT licensing, version 1.0.0, installation and environment-variable notes, layered references, examples, and best practices. Repository context also shows official-organization provenance, maintenance language, and CI/evaluation workflows. Points are deducted because the skill lacks a changelog, explicit maintainer and update path, dependency/version guidance including optional pandas, FAQs, troubleshooting, and sufficient limitation disclosure; some reference guidance is generic and not query-specific.

The evidence provides patterns for logs, metrics, batching, partial results, aggregations, dimension filters, metric definitions, and async clients, offering useful coverage for common coding tasks. Points are deducted because no skill-specific output or end-to-end result is statically verifiable, examples have missing imports and context dependencies, and production guidance for retries, paging/limits, authorization failures, and result completeness is insufficient.

The evidence includes the selected skill and reference files, version metadata, MIT license, and repository-level testing and CI signals. Points are deducted because the supplied tests do not cover the skill's key Python query paths, no skill-specific acceptance criteria, pinned dependencies, or third-party execution evidence are shown, and important API claims rely mainly on author-provided documentation rather than corroborating sources.