Agent Platform 1.1.3
The output of 'npm run prove' on this release, produced on 2026-10-29 by the machine that built the download.
There is no public repository yet. Ask us for the source and run this yourself: that is the point of the command.


Proving the controls of Agent Platform — 2026-10-29
Nothing here needs an account, an API key or a network connection.

  PROVEN   If the sandbox cannot start, the command does not run.
           19 test(s) · lib/exec-target.test.ts · claimed in SECURITY-OVERVIEW.md, How agents are contained
  PROVEN   An agent's shell reaches only the websites its team is allowed.
           30 test(s) · lib/agent-tools.test.ts · claimed in SECURITY-OVERVIEW.md, What an agent can reach, rung by rung
  PROVEN   A spend ceiling is checked while work runs, counts money not yet billed, and holds across parallel agents.
           35 test(s) · lib/limits.test.ts, lib/run-budget.test.ts · claimed in SECURITY-OVERVIEW.md, Money
  PROVEN   Anything written outside this app reaches a model fenced and labelled, never as an instruction.
           51 test(s) · lib/web.test.ts, lib/triggers.test.ts · claimed in SECURITY-OVERVIEW.md, How agents are contained
  PROVEN   The activity record is hash-chained: an entry cannot be changed or removed without the record saying so.
           21 test(s) · lib/audit.test.ts · claimed in SECURITY-OVERVIEW.md, The record
  PROVEN   A rehearsal changes nothing — no file written, no command run, nothing sent.
           57 test(s) · lib/procedure-run.test.ts, lib/plan-mode.test.ts, lib/email-rehearsal.test.ts · claimed in PROCEDURES.md, Rehearsal
  PROVEN   A team with nobody accountable for it runs nothing unattended.
           26 test(s) · lib/team-gate.test.ts, lib/schedules.test.ts · claimed in SECURITY-OVERVIEW.md, Who may do what
  PROVEN   Every email an agent sends says it was sent by an agent.
           48 test(s) · lib/messaging.test.ts, lib/smtp.test.ts, lib/email-policy.test.ts · claimed in SECURITY-OVERVIEW.md, The one regulation this already satisfies
  PROVEN   What a connector returns is fenced before the model reads it — a record in your CRM is data, not an instruction.
           14 test(s) · lib/connector-fence.test.ts · claimed in SECURITY-OVERVIEW.md, How agents are contained
  PROVEN   A connected app may only run the tools a person ticked, and a tool it did not have when they ticked is refused.
           37 test(s) · lib/connector-permissions.test.ts, lib/connector-tools-ui.test.ts · claimed in SECURITY-OVERVIEW.md, What an agent can reach, rung by rung
  PROVEN   What the permissions page says an agent may do is what the app asks before it lets the agent do it.
           22 test(s) · lib/capability-matrix.test.ts, lib/capability-matrix-ui.test.ts · claimed in SECURITY-OVERVIEW.md, Every agent, and what it may do

  The tests are checked too: 5655/5657 planted bugs caught, last full run 2026-10-27.
  (Each control is deliberately broken and the suite must fail: node tools/check-reliability-regressions.mjs)

  All 11 published claims proven on this machine.