Gauge Agents
Test changes before you merge
- GitHub PR checks. Test changes to your docs, Skills, and packages with coding agents using PR previews and CI artifacts. Gauge reports results in a PR comment and status check. Set up PR checks.
- Open-source Gauge CLI. Run coding agents locally to find where they struggle with your Skills, CLI, SDK, or docs, then test fixes before publishing. The full evaluate, analyze, fix, and retest loop runs with Docker and your own agent credentials—no Gauge account required. Get started on GitHub.
- Preference checks in Git. Write open-ended or head-to-head Agent Preference cases alongside your code. Results report what agents selected as advisory findings. Write preference cases.
Gauge Chat
Schedule work and connect more tools
- Scheduled Ask Gauge tasks. Schedule recurring Ask Gauge work from chat or Settings, run it on demand, and have scheduled tasks send results by email. Schedule a task.
- Slack messages from Ask Gauge. Ask Gauge and workflows can post messages to channels the Gauge bot belongs to. Send to Slack.
- Lock shared memory. Lock your organization’s memory to prevent edits from the Memory page, Ask Gauge, and MCP. Unlock it when you’re ready to make changes. Lock memory.
- More MCP tools. Sync Library from your sitemap to pick up pages published outside Gauge and match drafts to their live URLs. Query AI traffic and Gauge documentation directly from your AI client. Explore Gauge MCP.
Gauge Chat
Save workflows your team can reuse
- Reusable Skills. Save instructions for recurring jobs such as competitor reports or content refreshes. Run a Skill in Ask Gauge when you need it, or let Gauge choose one when your request matches. Create a Skill.
- More MCP tools. Create and manage your organization’s Skills, update keywords across multiple prompts, and publish content to a connected Sanity project from your AI client. Explore MCP tools.
Gauge Agents
Build and improve your agent evaluations
- Committed evaluation files. Define evaluation cases in Markdown and run them through the CLI, keeping tasks and criteria versioned with your repository. Use the CLI.
- AX Check. Run a public check of a developer product’s website, watch live coding sessions, and share a report with findings and recommendations. Run an AX Check.
- Generate evals from a repository. Repository-based eval generation is available to all organizations, giving you a starting set of tasks to review and run. Generate evals.
- Optimization workflows. Test improvements to documentation, Skills, and CLI behavior against recorded baselines. Propose edits across multiple pages and inspect results before adopting a change. Improve agent results.
Gauge Chat
See what AI searches for
- Fanouts. Explore the searches AI models issue behind the scenes, how they rewrite prompts, and when brands enter the research process. Export the search data for further analysis. Explore Fanouts.
- Custom traffic ingestion. Set up a custom log drain to send AI traffic events from infrastructure beyond the built-in hosting integrations. Set up ingestion.
- Share content with your team. Share a content link with teammates so they can open the same work in Gauge. Share content.
- SEO context for citations. A new SEO view on citation details adds context for evaluating a cited page. View SEO context.
- More MCP tools. Check a batch of article ideas against existing Library content before writing, helping your AI client identify overlap and avoid duplicating content. Check content coverage.
- Brand mention share in citations. Compare brand mention share from the Citations view. Compare brand coverage.
Gauge Agents
Run repeatable agent evaluations
- Agent Experience evals. Define a coding task with explicit pass criteria, compare results across run targets, and inspect the evidence behind each verdict. Create an eval.
- Authenticated MCP connections. Connect MCP servers through OAuth and select the credentials agents need during a run. Manage credentials.
- Agent Traffic. Inspect visits from AI agents in a dedicated dashboard, with Cloudflare and Vercel connection flows. Connect Agent Traffic.
- More agent and model choices. Run supported open models and agent harnesses through the public run workflow. Choose agents and models.
Gauge Agents
Bring Gauge into your development workflow
- CLI and public API. Manage measurements, launch runs, and inspect session logs, analysis, and code changes from the terminal or your own scripts. Get started with the CLI or explore the API.
- Private repositories. Connect the Gauge Agents GitHub App to give coding sessions access to selected private repositories. Connect a repository.
- Multi-model measurements. Select multiple models for a run or recurring schedule. Configure run targets.
- Configurable personas. Choose a simulated user’s perspective when starting a run to test different buyer contexts. Configure personas.
- Inspect agent research. Review the searches agents issued and the pages and domains they fetched during a coding session. Inspect a session.
Gauge Chat
Refresh existing content
- Refresh existing pages. Ask Gauge can attach a refresh to the existing Library page, keeping the work connected to the content you’re updating. Publish and measure content.
Gauge Chat
Share insights and publish from Gauge Chat
- PDF and PowerPoint exports. Ask Gauge can produce PDF reports and PowerPoint decks so you can share findings with clients and teammates. Create a report.
- Proactive Slack alerts. Receive new recommended actions in a channel you choose, with links back to Action Center. Alerts are grouped into a daily post. Connect Slack.
- GitHub publishing. Connect a repository containing Markdown content. Gauge opens a pull request for new or updated posts so your team can review and publish through its normal workflow. Connect GitHub.
- Sanity integration. Connect a Sanity project and dataset, import existing posts into Library, and publish content back to Content Lake. Connect Sanity.
- Use uploaded images in content. Ask Gauge can browse your uploaded images and embed them in content. Use your images.
- Update memory from chat. Ask Gauge can save information to your organization’s memory. Update memory.
- ChatGPT Ads exports. Export ChatGPT Ads data as CSV for reporting outside Gauge. Export your data.
- Content workflows over MCP. Connected AI clients can research, outline, write, review, and publish content through Gauge’s tools, with background job status and Action Center tracking. Publishing requires a connected CMS. Create content over MCP.
- Google Ads over MCP. Query account summaries, campaign reports, and keyword reports from connected Google Ads accounts through MCP. Use connected reporting tools.