Sanityops Inspect Lite — AI skill for Claude Code
Inspect-only, zero-install inspector for AI Agent logic artifacts (System Prompt, Skill, Tool Schema) - reports defects, cross-artifact consistency, and permission proportionality, with candidate Risk.
How to install Sanityops Inspect Lite
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open sanityops-org/sanityops-inspect-lite and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Sanityops Inspect Lite does
Inspect-only, zero-install inspector for AI Agent logic artifacts (System Prompt, Skill, Tool Schema) - reports defects, cross-artifact consistency, and permission proportionality, with candidate Risk/Quality impact mapping. A Claude Code plugin based on SanityOps Framework v1.0.
Alternatives in AI
- Geo SEO Claude — GEO-first SEO skill for Claude Code 3.6k ★
- NotebookLM Research Agent — A fully autonomous AI research agent that ingests sources into Google NotebookLM, runs deep web research, synt 168 ★
- Claude Ops Inspector — Subagent Verification for Claude AI Code Networks 2026 116 ★
README
title: SanityOps Inspect Lite — README description: "Orientation for the sanityops-inspect-lite inspector (SanityOps Framework v1.0): what to read first, where the skill sits among the framework's Inspect / Risk / Quality pillars, what it is and is not, the seven ordered inspection stages, lite-vs-full comparison, upgrade path, worked samples, and FAQ. Non-normative auxiliary guide."
SanityOps Inspect Lite — README
This file is the **auxiliary guide** for the `sanityops-inspect-lite` inspector. It covers the reading order, where the skill sits in the framework, what the seven ordered inspection stages are, how to use it, the lite-vs-full comparison, how to upgrade to the official tooling, worked samples, and FAQ. For the operational rules — inputs, the seven-stage flow, run modes, and the output contract — see [SKILL.md](plugins/sanityops-inspect-lite/skills/sanityops-inspect-lite/SKILL.md).
This is a **non-normative** document. It defines no new rules and never overrides the SanityOps Framework specification or the condensed checklists in `references/`.
What to read first
This package is read by two different audiences, so the entry point differs:
| Reader | Read first | Then |
|---|---|---|
| A human adopting, reviewing, or evaluating the package | this README — orientation: what it is, where it sits, what it does not do | SKILL.md for the rule-driven flow, then its references/ and samples/ |
| A host model executing an inspection | SKILL.md — the operational contract: inputs, the seven ordered stages, run modes, output contract |
the references/*.md checklists it points to, at the moment each stage runs |
The README is not a substitute for `SKILL.md`, and `SKILL.md` is not an introduction: the README answers *"should I use this, and what exactly is it?"*, while `SKILL.md` answers *"what must I do, in what order, and what must I emit?"*. A hu
Related Skills
Geo Review
GEO review for LLM-citation visibility — runs in diff mode (per-PR extraction checks) or site mode (cross-page
AI Engineering Claude C3 Agent Evaluation And Observability Project
Three Claude AI agent pipelines demonstrating evaluation and observability: validated retry with human-in-the-
Beckett Godot MCP
Zero-sidecar MCP server inside the Godot editor. AI agents (Claude Code, Cursor) inspect, author, run and SEE
AI Component Vetting
Vets third-party AI tooling components (MCP servers, hooks, agents, plugins, skills) for supply-chain risk, us
Striatum
A CLI for packaging and distributing AI artifacts (skills, prompts, workflows) as OCI images with dependency r
Pi Claude Artifacts
Pi extension: publish claude.ai artifacts via Claude Code's Artifact tool
Related Agents
Evidence Reviewer
Skeptical auditor of evidence bundles against .claude/skills/evidence-standards.md. Detects circular citations
Of Reviewer
OwnFramework Loop reviewer — inspect one exact candidate SHA and fill one pass-scoped semantic assessment. Rea
Oracle Evaluator
Runs the evaluation harness on candidate artifacts via pp_harness best-of-N and Borda. Invoke after a draft ha