sanity-io/agent-context
3 skills1.4k installs12 starsGitHub
Install
npx skills add https://github.com/sanity-io/agent-contextSkills in this repo
1Create Agent With Sanity Contextcreate-agent-with-sanity-context is a skill from sanity-io/agent-context for building AI chatbots and assistants powered by Sanity CMS data. The skill covers Sanity Studio setup, agent implementation with client-side tools, and advanced patterns for connecting Claude, GPT, or other LLMs to structured content. Developers reach for create-agent-with-sanity-context when they need semantic search over Sanity data, schema-aware content retrieval, or a Sanity Context MCP server integration rather than raw embedding lookups. The skill addresses chatbot setup, AI assistant configuration, and making CMS content searchable by agents with proper schema context.522installs2Shape Your Agentshape-your-agent is an optional, interview-style skill from sanity-io/agent-context for developers who control their agent's system prompt while using the Sanity Context MCP server. It walks through use-case discovery, concrete tone and verbosity choices, boundary rules with trigger scenarios, and a 200-400 word prompt draft that avoids duplicating GROQ, schema, or formatting guidance already handled by the MCP and Instructions field. The workflow includes a cut test for untriggerable rules, separation tables for system prompt versus Instructions field concerns, and iteration against five or more real user questions. Reach for shape-your-agent when an agent sounds wrong, refuses poorly, or needs communication policies beyond what dial-your-context configures for Sanity data access.445installs3Dial Your Contextdial-your-context is a Sanity-io agent-context skill for interactively authoring Instructions field content on the Sanity Context MCP server. The session produces concise pure deltas containing only information the base model lacks, such as schema relationship guidance, content filters, and data-specific query rules. Developers reach for dial-your-context when agents return wrong Sanity query results, confuse schema relationships, or need optimized content access boundaries. Triggers include tuning agent context, configuring MCP instructions, improving agent responses to Sanity data, and fixing agents that misunderstand which content they can access.443installs