ESTIMATOR
Agent risk estimator: match agentic endpoint controls to the agents you run
SHORT ANSWER
Tell the estimator which agents your people run, where they run and what they are allowed to do, then set how much each criterion matters. It counts your exposure on a plain checklist, shows which products name your agents and document the verdicts you need, and re-ranks all thirteen scored products on your weights. Scores are editorial; nothing here measures your environment.
What is your exposure?
Your agents can do 0 of the 6 things on our checklist.
Narrow
This is a checklist count, not a measured risk score. It tells you which criteria to weigh more, nothing about your environment.
Which products fit your weights?
| Rank | Product | Re-weighted score | Published weights | Your agents named | Your verdicts documented | Cloud reach |
|---|---|---|---|---|---|---|
| 1 | Bay | 7.20 / 10 | 7.20 · rank 1 | 1 of 3 Not named: Cursor and GitHub Copilot | 1 of 1 | 3 / 10 |
| Tied T-2 | Neo | 6.85 / 10 | 6.85 · rank T-2 | 2 of 3 Not named: Cursor | 1 of 1 | 5 / 10 |
| Tied T-2 | Onyx Security | 6.85 / 10 | 6.85 · rank T-2 | 0 of 3 Not named: Claude Code, Cursor and GitHub Copilot | 1 of 1 | 9 / 10 |
| 4 | Zenity | 6.75 / 10 | 6.75 · rank 4 | 3 of 3 | 1 of 1 | 9 / 10 |
| 5 | Noma Security | 6.55 / 10 | 6.55 · rank 5 | 2 of 3 Not named: GitHub Copilot | 1 of 1 | 9 / 10 |
| 6 | Lasso Security | 6.40 / 10 | 6.40 · rank 6 | 2 of 3 Not named: GitHub Copilot | 1 of 1 | 6 / 10 |
| 7 | CrowdStrike Falcon Guardian | 6.25 / 10 | 6.25 · rank 7 | 1 of 3 Not named: Cursor and GitHub Copilot | 1 of 1 | 8 / 10 |
| 8 | Harmonic Security | 5.90 / 10 | 5.90 · rank 8 | 2 of 3 Not named: Cursor | 1 of 1 | 4 / 10 |
| 9 | Ent | 5.80 / 10 | 5.80 · rank 9 | 3 of 3 | 1 of 1 | 3 / 10 |
| 10 | Koi (Palo Alto Networks) | 5.75 / 10 | 5.75 · rank 10 | 2 of 3 Not named: GitHub Copilot | 1 of 1 | 6 / 10 |
| 11 | Prompt Security (SentinelOne) | 5.45 / 10 | 5.45 · rank 11 | 3 of 3 | 1 of 1 | 6 / 10 |
| 12 | Glow Security | 5.40 / 10 | 5.40 · rank 12 | 1 of 3 Not named: Cursor and GitHub Copilot | 1 of 1 | 4 / 10 |
| 13 | Bloom Security | 4.65 / 10 | 4.65 · rank 13 | 0 of 3 Not named: Claude Code, Cursor and GitHub Copilot | 1 of 1 | 2 / 10 |
Formula: re-weighted score = sum(criterion score × your weight) / sum(your weights). Criterion scores are our published editorial scores from 0 to 10, from vendor pages read on 27 September, 11 October and 13 October 2026. No vendor publishes a per-device or per-user price, so the estimator shows no costs; the AWS Marketplace list prices for Noma Security and Lasso Security are on their profiles.
With your agents (Claude Code, Cursor and GitHub Copilot), 3 of 13 products name all of them; 13 of 13 document every verdict you checked.
Frequently asked questions
Does the estimator measure my risk?
No. The exposure tally counts the items you checked, and the ranking re-weights our editorial scores. Nothing scans your devices or agents.
Why does a product not move when I uncheck an agent?
Agents and verdicts change the columns and the filters, not the scores. Only the weight sliders change the ranking.
Why are there no prices?
No vendor publishes a per-device or per-user price. Noma Security and Lasso Security list annual contracts on AWS Marketplace, shown on their profiles. We do not estimate prices.
What do the suggested weights do?
They raise the weight of stopping actions before they run, cloud and workspace reach or verdict range when your answers point that way, and say which sliders changed. You can move any slider afterwards.
Scores are editorial assessments from public sources read on the date shown.