CAIN-42 MCPGateAGENTIC AI 75 · #59
More efficient browser representations and action loops
Engineering evaluations examined trimmed accessibility trees, action batching and adaptive waiting.
Last reviewed 2026-11-11
Agentic AI 75 / #59
2026 status: Engineering evaluation · Category: Browser agents
The 2026 development
Engineering evaluations examined trimmed accessibility trees, action batching and adaptive waiting.
The governance gap
Efficiency changes the failure mode. Batching actions reduces the number of checkpoints, which can turn several independently reviewable steps into one opaque leap.
Where CAIN-42 fits: Moat 5 — Predictive Trust & Blast Radius
Core question: What could happen before we let it happen? Pre-execution analysis of reach, exposure and consequence, including simulations.
Keep the checkpoint at the consequential boundary, not at every keystroke. Batching is allowed for reversible preparation; an irreversible action still needs its own decision.
How the wiring works
- Classify batched actions as reversible or consequential.
- Allow batching of reversible preparation.
- Require a separate checkpoint before an irreversible effect.
- Record the expanded action sequence for review.
What is real today
LIVE A public route in this deployment exercises this mechanism today, within its stated scope.
The browser firewall demo evaluates fixed presets. It does not measure browser-agent efficiency.
| Component in this repository | What the module says it is |
|---|---|
cain/browser/agent_firewall.py | Browser Agent Action Firewall. |
cain/argus/path.py | CAIN ARGUS — execution-path assurance (Prompt 14, section 7). |
platform-gateway/action_risk.py | Action risk: how dangerous is the action itself, independent of how its text reads. |
Descriptions are the first line of each module's own docstring, read without importing it. A module path is a pointer into this repository, not a hosted endpoint. Its existence is checked by the test suite; its status for this use is the label above.
Check it yourself
- Run the live browser-agent demo
/api/v1/browser-agent-firewall/presets - Fire a live decision through the CAIN-42 pipeline
/fabric/try— the same enforcement path every product sits behind.
Commercial package
| Offer | Browser Agent Control Review |
|---|---|
| Buyer | Browser-agent vendors and web platforms |
| Price (proposal) | $2,500–$5,000 per review |
| Bundle | Commerce & Payments — Delegated spending with a ceiling and a receipt. |
| Operator stream | Stream 2 of the 10-stream operator playbook |
Fintech, commerce platforms, marketplaces, treasury teams is the bundle buyer. The entry offer for this bundle is Payment-mandate review; browser-agent checkout review; the range is $2,500–$7,500 per review; integration from $10,000. A price on this page is a proposal, not a validated rate, and no customer is claimed.
Buy this as a product or service
CAIN-A75-059 — Browser Agent Control Review PILOT
A bounded pilot report: up to five scenarios, the relevant library engine or analysis, a written result and the stated limits. Engineering time is quoted per pilot.
Kind: Assessment · Price (proposal): $2,500–$5,000 per review · Lead time: 10 business days.
Open the product page · all 75 products
What this page does not claim
What this page is not. CAIN-42 is not affiliated with, endorsed by, integrated into, or shipped with the organisation named in the 2026 item. The item is public context; the page describes how CAIN-42 would govern that class of capability. No page here claims a customer, a partnership, a certification, a patent, or a production integration. Prices are proposals. Built by a three-person team.