Visual testing just became the guardrail for agentic code. applitools shipped visual ai guardrails…
visual testing just became the guardrail for agentic code. applitools shipped visual ai guardrails on september 15, catching quality regressions as ai writes the changes, with eyes mcp tools that never send your ui to public models.
the agent ships the diff, the guardrail checks the look.
Context
Applitools' release of September 15, 2026, republished by AiThority, announces a platform release with three items: Eyes Visual AI MCP tools (@applitools/mcp), which connect Visual AI to agents such as Claude Code, Cursor, Copilot and Cline to pinpoint visual differences and maintain baselines, Figma Design Baselines, and NLP Test Steps for SDKs on a proprietary Deterministic Language Model. It frames this as a deterministic governance layer for the agentic SDLC.
Everything here is vendor-reported. The privacy line, that sensitive UI data is never sent to public foundation models or used to train third-party AI, is the release's statement of SOC 2 Type II processing and is not independently verified. Whether any call path touches a third-party model was not checked. The guardrails framing is the release's. The release cites Faros AI telemetry on review times and incident risk, which was not inspected and is not used.
Related work
- The pixel is the last unchecked artifact in agentic coding ↗The earlier note on the same September 15 Eyes MCP tools release. This one adds the guardrails framing and the privacy claim.
Watch next
- Applitools docs on MCP data handling, and independent checks of baseline maintenance accuracy.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 22 September 2026 at 01:32 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →