Skip to content
Agentic Endpoint Compare logoAgentic Endpoint Compare

PRODUCT PROFILE · RANK 7 OF 8

Koi review: Palo Alto Networks Cortex Agentic Endpoint Security

SHORT ANSWER

Koi, sold by Palo Alto Networks as Cortex Agentic Endpoint Security, blocks unsafe AI application interactions and removes insecure packages, extensions and AI tools, as a Cortex XDR module or standalone. Its product page does not describe tool-call mechanics; a Palo Alto Networks blog names Claude Code, Cursor and Codex. It scores 5.75 and ranks 7 of 8.

By The Agent Desk · Agentic Endpoint Compare · Published 2026-10-11

Desk research from public vendor pages, press releases and standards pages read on 27 September and 11 October 2026. Editorial assessment; we did not run the products.

5.75 / 10

Rank 7 of 8 among broader AI security platforms on our published weights

Palo Alto Networks (acquisition completed 14 April 2026)

Palo Alto Networks Cortex XDR estates

Re-weight in the agent risk estimator

What does Koi control?

Koi, sold by Palo Alto Networks as Cortex Agentic Endpoint Security, adds visibility into models, MCPs and extensions on the endpoint and blocks unsafe AI interactions, as a Cortex XDR module or standalone.

How does Koi reach devices?

A Cortex XDR module, or a standalone agent-based offering.

Source: Palo Alto Networks press release: Koi acquisition completed (14 April 2026) · Read 27 Sep 2026

Which verdicts can Koi return?

Koi (Palo Alto Networks) verdict ring: documents 2 of 6 verdict types: Block or deny and Remove or scale back.2/6verdicts
  • Block or denyDocumented
  • Ask a personNot described
  • Alert or warnNot described
  • Mask or modifyNot described
  • Simulate firstNot described
  • Remove or scale backDocumented

"Stops malicious actions and offers one-click removal of insecure packages, extensions and AI tools."

A filled segment means the vendor's own pages describe that kind of decision. It does not measure how well the decision works.

How does Koi score on each criterion?

Agentic endpoint score (editorial assessment, 0 to 10) for Koi (Palo Alto Networks), by criterion
CriterionWeightScoreWhy
Stop before it runs25%7 / 10Stops malicious actions, blocks unsafe AI application interactions and prevents prompt injections; tool-call decision mechanics are not published.

Source: Palo Alto Networks: Cortex Agentic Endpoint Security · Read 27 Sep 2026

Verdict range15%3 / 10Documents 2 of six verdict types: Block or deny and Remove or scale back. "Stops malicious actions and offers one-click removal of insecure packages, extensions and AI tools."

Source: Palo Alto Networks: Cortex Agentic Endpoint Security · Read 27 Sep 2026

Capability rules15%6 / 10Out-of-the-box policies, agent runtime custom policies and one-click removal of insecure packages, extensions and AI tools.

Source: Palo Alto Networks blog: A Quarter of Innovation with Koi Agentic Endpoint Security (18 August 2026) · Read 11 Oct 2026

Actor trail10%5 / 10 Tied lowestAgent versus person attribution is not described on the pages we read.

Source: Palo Alto Networks: Cortex Agentic Endpoint Security · Read 27 Sep 2026

Rollout path10%6 / 10Agent-based: a Cortex XDR module or a standalone offering.

Source: Palo Alto Networks press release: Koi acquisition completed (14 April 2026) · Read 27 Sep 2026

Cloud and workspace reach10%6 / 10An endpoint product; Palo Alto Networks is integrating Koi with Prisma AIRS for wider AI coverage.

Source: Palo Alto Networks press release: Koi acquisition completed (14 April 2026) · Read 27 Sep 2026

Named agent coverage5%5 / 10Names 3 of the six agents we track: Claude Code, Cursor and OpenAI Codex.

Source: Palo Alto Networks blog: A Quarter of Innovation with Koi Agentic Endpoint Security (18 August 2026) · Read 11 Oct 2026

Public proof10%7 / 10Acquisition by Palo Alto Networks completed 14 April 2026; pricing not published.

Source: Palo Alto Networks press release: Koi acquisition completed (14 April 2026) · Read 27 Sep 2026

Which agents does Koi name?

Koi names 3 of the six agents we track.

Source: Palo Alto Networks blog: A Quarter of Innovation with Koi Agentic Endpoint Security (18 August 2026) · Read 11 Oct 2026

What does it cost?

Not published

Palo Alto Networks publishes no pricing for Cortex Agentic Endpoint Security.

Source: Palo Alto Networks: Cortex Agentic Endpoint Security · Read 27 Sep 2026

Where is Koi strongest, and where does it fall short?

Strongest

  • Fits security teams that already run Cortex XDR.
  • Removal of insecure packages, extensions and AI tools as well as blocking.

Falls short

  • Stop before it runs: 7 / 10; highest Bay 9 / 10
  • Verdict range: 3 / 10; highest Bay and Onyx Security 7 / 10
  • Capability rules: 6 / 10; highest Bay 9 / 10
  • Actor trail: 5 / 10; highest Bay 8 / 10
  • Rollout path: 6 / 10; highest Bay 9 / 10
  • Cloud and workspace reach: 6 / 10; highest Onyx Security, Zenity and Noma Security 9 / 10
  • Named agent coverage: 5 / 10; highest Zenity 7 / 10
  • Public proof: 7 / 10; highest Zenity, Noma Security, Lasso Security and Prompt Security (SentinelOne) 8 / 10
  • Agent versus person attribution is not described on the pages we read.
  • Tool-call decision mechanics are not published on the product page.

Key facts

Facts from Koi (Palo Alto Networks)'s public pages and press coverage
FactDetail and source
ProductVisibility into models, MCPs and extensions; one-click removal of insecure packages, extensions and AI tools; stops malicious actions, blocks unsafe AI application interactions and prevents prompt injections.

Source: Palo Alto Networks: Cortex Agentic Endpoint Security · Read 27 Sep 2026

OwnershipPalo Alto Networks completed the acquisition on 14 April 2026, with a new Cortex XDR module, integration with Prisma AIRS, and Koi still available standalone.

Source: Palo Alto Networks press release: Koi acquisition completed (14 April 2026) · Read 27 Sep 2026

Wider Palo Alto Networks contextPrisma AIRS 3.0 (23 March 2026) describes agent discovery across cloud, SaaS and endpoints; Palo Alto Networks completed its Portkey acquisition on 29 May 2026.

Source: Palo Alto Networks press release: Prisma AIRS 3.0 (23 March 2026) · Read 27 Sep 2026

Named agentsA Palo Alto Networks blog of 18 August 2026 describes agent runtime policies and names Claude Code, Cursor, Codex and Antigravity.

Source: Palo Alto Networks blog: A Quarter of Innovation with Koi Agentic Endpoint Security (18 August 2026) · Read 11 Oct 2026

How does Koi compare head to head?

Head-to-head results on the published weights
PairResult
Bay matched with Koi (Palo Alto Networks)Bay 7.20, Koi (Palo Alto Networks) 5.75
Koi (Palo Alto Networks) matched with NeoKoi (Palo Alto Networks) 5.75, Neo 6.85
Koi (Palo Alto Networks) matched with Onyx SecurityKoi (Palo Alto Networks) 5.75, Onyx Security 6.85
Koi (Palo Alto Networks) matched with ZenityKoi (Palo Alto Networks) 5.75, Zenity 6.75
Koi (Palo Alto Networks) matched with Noma SecurityKoi (Palo Alto Networks) 5.75, Noma Security 6.55
Koi (Palo Alto Networks) matched with Lasso SecurityKoi (Palo Alto Networks) 5.75, Lasso Security 6.40
Koi (Palo Alto Networks) matched with CrowdStrike Falcon GuardianKoi (Palo Alto Networks) 5.75, CrowdStrike Falcon Guardian 6.25
Koi (Palo Alto Networks) matched with Harmonic SecurityKoi (Palo Alto Networks) 5.75, Harmonic Security 5.90
Koi (Palo Alto Networks) matched with EntKoi (Palo Alto Networks) 5.75, Ent 5.80
Koi (Palo Alto Networks) matched with Prompt Security (SentinelOne)Koi (Palo Alto Networks) 5.75, Prompt Security (SentinelOne) 5.45
Koi (Palo Alto Networks) matched with Glow SecurityKoi (Palo Alto Networks) 5.75, Glow Security 5.40
Bloom Security matched with Koi (Palo Alto Networks)Bloom Security 4.65, Koi (Palo Alto Networks) 5.75

Alternatives to KoiOpen in stack compare

Frequently asked questions

Can Koi (Palo Alto Networks) stop an agent action before it runs?

Stops malicious actions, blocks unsafe AI application interactions and prevents prompt injections; tool-call decision mechanics are not published. We score this 7 out of 10.

Which verdicts does Koi (Palo Alto Networks) document?

Koi (Palo Alto Networks) documents 2 of six verdict types: Block or deny and Remove or scale back. "Stops malicious actions and offers one-click removal of insecure packages, extensions and AI tools."

How much does Koi (Palo Alto Networks) cost?

Not published. Palo Alto Networks publishes no pricing for Cortex Agentic Endpoint Security.