The free three week pilot for security teams running Claude Code agents. It opens in transparent mode, with zero developer friction. At the end of week one, Interpose hands your team a report of what your agents actually did. Then block mode, at the pace you set.
On-prem, in your own boundary. Scoped to one live use case. Nothing reaches Tuent.
In transparent mode Interpose takes no action. Only the default credential floor blocks. Everything else is watched and written to the signed, tamper-evident trail while your agents keep working. At the end of the week you get one report: the tools called, the files read, the targets touched, and anything that broke the agent's own pattern. Most teams have never seen this. The report is yours to keep whether or not you buy.
When your team has seen enough, you move controls to block mode. Out-of-bounds actions are then stopped at the pre-execution gate, before the side effect lands. The decision is deterministic, with no model in the block path. Every control has an alert setting and a block setting, so you decide which risks you watch and which you stop, one at a time.
We sit down with your team and shape the use case from the week one data. Every finding in the report becomes a decision: watch it, or stop it. Interpose drafts the rule, your team approves it, and each control moves from alert to block one at a time, at the pace you set. Developers still change nothing.
Block mode runs live on the use case. Denials land at the pre-execution gate, deterministically, with no model in the block path. When a block is wrong, the operator restores the agent in one click and the rule tightens from the miss. By the end of the week your team has watched Interpose stop real behavior on real work.
We close the pilot with the full record: what was watched, what was stopped, and what the policy now covers. The report is yours to keep whether or not you buy. If you continue, a signed letter of intent locks your production rate at $40 per endpoint per month before the fleet grows.