@zedtech/zedkit-ai
Write each tool's agent files from one source and a policy
Generate `.cursor/agents`, `.claude/agents`, and Muse-native files in `.agents/agents` from the unmarked/kit source and `.agents/tokenomics.md`, with the model…
At a glance
- Who: Maintainers who keep subagents in
.agents/agentsand want each tool's copy to carry the model from one policy file. - What: Generate
.cursor/agents,.claude/agents, and Muse-native files in.agents/agentsfrom the unmarked/kit source and.agents/tokenomics.md, with the model of each agent's default route and a marker so a later run can tell its own files from yours; never overwrite a file it did not write. - Use when: After editing an agent or
tokenomics.md, after upgrading the kit, and as a--checkstep in CI. - Avoid when: When every agent runs on the session model and no tool needs its own copy. Codex reads
.agentsdirectly. Muse gets a native marked file in.agents/agents(not the canonical source). - See also: Route an agent call to a model and effort from a policy file
Install
pnpm add @zedtech/zedkit-aiLatest: 1.2.0 (via changesets).
Example
import { inSync, syncKit } from "@zedtech/zedkit-ai";
const actions = syncKit({
dest: ".",
providers: ["cursor", "claude"],
check: true,
});
if (!actions.ok) throw actions.error;
if (!inSync(actions.value)) process.exitCode = 1;Wraps: (none)
Route an agent call to a model and effort from a policy file
Resolve kind (how the agent is used for this call) times stakes (how much is at risk) and the call's needs to a provider's model and effort, with the gates of…
Show the logo in a React app
Render the logo as an inline SVG that follows the text colour, so light and dark need no second asset.