LogoReasonable Application Security
Authors
Subscribe
LogoReasonable Application Security

Archive

The approval saved the password.

Nov 17, 2026

•

3 min read

The approval saved the password.

What “don’t ask again” writes to disk, Coinbase’s server-side agent limits, JPMorgan’s CISO on leaving the open internet, five-minute secret abuse, hijacked Optimizely scripts, agents that turn patches into exploits, and Podcast Corner.

Chris Romeo
Chris Romeo
Your agent can read the approval code.

Nov 12, 2026

•

3 min read

Your agent can read the approval code.

Where your agents’ approvals land, archived MCP servers, poisoned tool updates, SAST vs. AI review, credentials stolen through CSP, AlphaProof Code, and Podcast Corner.

Chris Romeo
Chris Romeo
Auto-approve is a security setting.

Nov 5, 2026

•

3 min read

Auto-approve is a security setting.

What auto-approve costs your coding agents, agents that switched off their own monitoring, an AGENTS.md in Downloads, and Podcast Corner.

Chris Romeo
Chris Romeo
AI won’t wait for your security review.

Oct 27, 2026

•

4 min read

AI won’t wait for your security review.

Continuous security review, AI-driven break-ins, formal verification, and Podcast Corner.

Chris Romeo
Chris Romeo
The agent fixed the bug. The flaw moved one route over.

Oct 19, 2026

•

4 min read

The agent fixed the bug. The flaw moved one route over.

The patch was right. The export endpoint was still open. Plus shadow MCP servers and the AI Security Table.

Chris Romeo
Chris Romeo
A clean AI review still needs evidence.

Oct 13, 2026

•

5 min read

A clean AI review still needs evidence.

Turn AI security conclusions into checks your team can inspect. Plus Brook Schoenfield, AI Security Table, and the latest shorts.

Chris Romeo
Chris Romeo
Your coding agent is another privileged identity.

Oct 7, 2026

•

5 min read

Your coding agent is another privileged identity.

Claude and Codex permissions through an AppSec lens: identity, least privilege, execution boundaries, and evidence.

Chris Romeo
Chris Romeo
The Agents Took Over My Asana

Sep 29, 2026

•

3 min read

The Agents Took Over My Asana

What an issue-tracker experiment reveals about agent trust boundaries, tool permissions, verification, and runtime monitoring.

Chris Romeo
Chris Romeo
The Prompt Is Not the Trust Boundary: Reasonable Application Security #81

Sep 22, 2026

•

4 min read

The Prompt Is Not the Trust Boundary: Reasonable Application Security #81

Prompt injection is no longer just a bad prompt. Agents can find instructions anywhere they can read.

Chris Romeo
Chris Romeo
A Configured Exporter Is Not a Kill Switch

Sep 15, 2026

•

4 min read

A Configured Exporter Is Not a Kill Switch

A controlled experiment found two observability paths with different switches, payloads, and privacy profiles.

Chris Romeo
Chris Romeo
AI Made Code Cheap. Review Is the New Bottleneck

Sep 8, 2026

•

5 min read

AI Made Code Cheap. Review Is the New Bottleneck

Chris’s take, worthwhile security reads, and the latest podcast episodes.

Chris Romeo
Chris Romeo
The Agent’s Permissions Are the Security Boundary

Sep 1, 2026

•

3 min read

The Agent’s Permissions Are the Security Boundary

A practical control model for coding agents: identity, least privilege, constrained execution, and evidence you can review.

Chris Romeo
Chris Romeo
123...8
Reasonable Application Security

Reasonable Application Security

Striving to make application security reasonable.


Home

Posts

Authors

© 2026 Reasonable Application Security.
beehiivPowered by beehiiv