@guren/plugin-ai has 395 recorded weekly downloads on npm, an increase of 20.4% over the past 30 days.
It ranks #1519 of 2,251 tracked npm listings by weekly downloads.
PluginStats has 121 daily observations since Jun 19, 2026; the latest is from Oct 17, 2026.
About @guren/plugin-ai
@guren/plugin-ai
In-process AI agents for Guren applications (RFC 0029), on the Vercel AI SDK. An Agent calls a model configured in config/ai.ts and reaches the application only through appTools(), which runs every tool call through the same invocation pipeline as the App MCP endpoint: scopes, the route's policies, approvals, redaction and the audit trail.
With guren codegen run, .guren/agents.gen.ts types appTools(): a name no route derives is a compile error, each tool's input and result are typed from its route contract, and the Agent parameter makes a name missing from a tool: entry in scopes one too. A prefix grant (tools:tickets.*) is checked when as() constructs the agent.
A tool defined with tool() inside tools() runs with whatever authority its closure has, and nothing gates or audits it. Anything a route already does belongs in appTools().
settings sends AI SDK call settings with every model call (maxOutputTokens, providerOptions such as Anthropic's effort or cacheControl). continueWhen = isPausedTurn lets prompt() resume a turn Anthropic paused with pause_turn during server-side web search, up to maxContinuations (5) extra calls. A response carries sources, providerMetadata, modelId and rawFinishReason beside text, output, steps and usage, and computeCostUsd(usageOf(response), pricing) prices it, server-side tool calls included through pricing.perThousandRequests.
Tool names reach the model provider verbatim. Anthropic and OpenAI accept only [A-Za-z0-9_-]{1,64}, so a route named tickets.show needs .agent({ toolName: 'tickets_show' }).
In tests, app.fakeAi() from @guren/testing scripts the model per agent while the tools still run through the pipeline. It needs ai installed beside @guren/plugin-ai, since the fake is built on MockLanguageModelV4 from ai/test:
A prompt with nothing scripted fails when ai is disposed, naming the agent, even if the route turned the error into a 500.
Requires Node 22 or later (the AI SDK's floor), or Bun.
Description from the npm listing.
- guren
- bun
- typescript
- fullstack
- framework
- ai
- ai-agents
- ai-sdk
- llm
Where @guren/plugin-ai sits in the listing
Each band is one page of 20. Weekly downloads on a log scale. Placement is projected from weekly downloads, so it can differ from the marketplace's own sort order. Hover a dot to see the listing.
Charts
Images: 30d · 1y · all time · 1600×900 dark
Adoption over time
7/30/90-day growth
Filled points are observed values; hollow points come from Internet Archive snapshots of the listing page. Dashed segments span periods with no observations — values between points are not interpolated.
Recent observations
Full history (121)Similar npm AI SDKs plugins
Data source
Weekly downloads (trailing 7 complete days) come from the official npm downloads API. Downloads include CI and mirrors and are not equivalent to installs or users. npm publishes no ratings.
Listing: https://www.npmjs.com/package/@guren/plugin-ai. First seen Oct 16, 2026; last successful collection Oct 18, 2026. See the methodology.