From the counter
52 posts
-
Skill scanning in Claude Enterprise skips API uploads, pre-existing skills and ZDR orgs: closing the gaps
Claude Enterprise skill scanning checks new uploads and edits in claude.ai and Cowork. Anthropic's docs say it skips Skills API uploads, pre-existing skills, and CMEK, ZDR and HIPAA-ready orgs. A control for each gap.
-
Prove your agent policy fires: a harmless canary suite for Claude Code hooks and MCP allowlists
Several hook and allowlist settings fail open without an error. Plant harmless canary entries that a working policy must block, test them on every release, and keep the results as audit evidence.
-
Are built-in agent protections enough? What auto mode, MCP allowlists, store scanning and skill scanning cover
Native agent controls are real, and each covers its own surface: one client, one store, one upload path. The residual risk sits across vendors, across registries and after approval. A coverage map with the blank cells left blank.
-
Thirty minutes with the board: briefing directors on AI coding agents without fear-selling or false comfort
Directors give CISOs about 30 minutes and want better AI-risk reporting. Brief them on coding agents with four slides - one exposure number, one incident pattern, one control decision, one ask - each built from a sourced figure.
-
The axios compromise from an agent operator's seat: which controls would have fired, and when
Replaying the 31 March 2026 axios timeline against seven controls: a release-age delay and install-script blocking covered the whole window, a provenance check caught one of two versions, and advisories came at removal.
-
A classifier that misses 17% of overeager actions: layering model-based approval over deterministic deny rules
Anthropic's auto-mode classifier misses 17% of real overeager actions in its own evaluation. It is a good middle layer. Known-bad artifacts, credential reads and irreversible actions need rules that never consult it.
-
Approval prompts are not a control when people approve 93-97% of them: what Anthropic's data means for policy
Anthropic's own studies show users approve 93-97% of permission prompts and caught 13.6% of planted dangerous commands. Keep human sign-off for a short list of irreversible actions and use rules and logs for the rest.
-
When agent-skill scanners disagree: what conflicting verdicts on the same skill mean for your install policy
In one census, auditors gave the same skill HIGH or CRITICAL and LOW or SAFE ratings on 1,233 listings, and 87.8% of listings had no audit at all. Treat conflict and absence of audit as separate policy states, not as a pass.
-
Why agent-security statistics disagree: a source-lineage audit of the MCP and skills numbers vendors quote
We traced ten widely quoted MCP and skills statistics to their origins. Most measure different things on different populations; two have no traceable source. A register with a safe-to-cite verdict for each.
-
When an agent causes the loss and there's no attacker: the cyber-insurance trigger gap and what to ask your broker at renewal
Cyber policies usually respond to attacks and breaches. An authorised agent that deletes data or sends it to the wrong place may fit neither. Ask your broker, in writing, which clause would respond.
-
Your agent installed a known-bad package: the first hour, including the credentials you forgot it could reach
Isolate the machine, preserve the agent's logs and configs, rotate every secret the agent could read from a clean device, check memory files and persistence, then scope. A minute-by-minute runbook.
-
Agent-governance maturity model: five observable stages from 'anything installs' to 'enforced, logged, tested'
A five-stage model for AI coding-agent governance across six dimensions, where every stage is defined by evidence an auditor can inspect. Your stage is your weakest dimension, which makes it hard to game.
