sebbi.pro
TEMPORAL STANDING

Valid on paper isn't valid now.

An AI agent was allowed to pay a supplier yesterday. This morning the supplier's account was flagged for fraud. Nobody cancelled the permission. Should the payment go through? On sebbi.pro it doesn't. Every action is checked at the moment it happens against the facts its permission depends on, and if one has changed, the action is stopped and the refusal is sealed as evidence.

The full Temporal Standing Test, run on production

The Temporal Standing Test is a published protocol for whether a system really re-establishes authority at the moment an action takes effect. It's run here against the live engine, with the implementation frozen and sealed before every run, the four cases in random order, and every run listed, failures included.

Latest run r_71c8a4622a0c4bbc · 2026-11-02 11:07:23 · sealed in block 30303

PASS
CaseChange at dNRequired at TnResult
Arevocation, standing-defeatingDENY / NON-EXECUTABLEPASS
Brevocation, standing-preservingPERMIT / EXECUTABLEPASS
Cexternal fact, standing-defeatingDENY / NON-EXECUTABLEPASS
Dexternal fact, standing-preservingPERMIT / EXECUTABLEPASS

Full evidence package Every run (1)

Run it yourself, now. The test runs on production and the result is sealed whatever it is. One run every 10 minutes.

A · Revoked in the lineage

The permission is cancelled after approval. The action must not happen.

B · Revoked elsewhere

A different, unrelated permission is cancelled. The action must still happen, exactly once.

C · A fact it depends on changes

The permission is untouched: not revoked, not expired, not altered. But the fact it depends on changes. The action must not happen.

D · A fact it doesn't depend on changes

Some other fact changes. The action must still happen, exactly once.

Use it on your own agents

Attach the facts a permission depends on. They're checked at the moment every action binds, and they only ever narrow authority.

POST https://sebbi.pro/x/conditions/fact      {"fact": "your-bank/payee-p77", "value": "verified"}
POST https://sebbi.pro/x/conditions/attach    {"grant": "<grant id>", "conditions": [
   {"type": "fact",   "fact": "your-bank/payee-p77", "op": "eq", "value": "verified"},
   {"type": "switch", "id": "JG-XXXX-XXXX"},
   {"type": "url",    "url": "https://registry.example/licence/123", "pointer": "/status", "op": "eq", "value": "active"}]}
GET  https://sebbi.pro/x/conditions/check?grant=<grant id>

Kinds: a fact in the sebbi.pro fact register (written only by the account that owns its namespace, every change sealed), a human kill switch that must be live, or a value at a public https address. An unknown or unreachable fact never counts as holding. MCP tools: sebbi_standing_fact, sebbi_standing_attach, sebbi_standing_check.

Start free Agent Passport Kill switch

Protocol: Temporal Standing Test. Runs of the earlier revocation-only test stay at https://sebbi.pro/x/standing/runs.

AI news
The VergeThe biggest modding site in PC gaming is taking a stance on AI4hBBC NewsGraduates are managing AI - but are they learning much else?5hThe VergePlease stop asking me when the AI bubble will pop5hTechCrunchAnthropic is working on a system to prove AI models’ identity6hTechCrunchTed Schlein’s Ballistic Ventures seeks $490M for new funds amid AI-driven cybersecurity boom7hThe VergeApple starts giving Siri AI access to third-party apps9hTechCrunchAI inference startup Fal doubles its valuation to $16B as demand surges9hTechCrunchNetflix’s new AI Discover tab helps you find something to watch based on your mood10hThe VergeOpenAI says ChatGPT is catching up to Google in search11hThe VergeNetflix will show you themed recommendations through a new ‘AI Discover’ tab11hArs TechnicaGoogle will let schools skip AI tools, but they have to give up search, too11hTechCrunchAnthropic is still grappling with the economics of compute12hWiredHere’s Why You Shouldn’t Use AI to Vote14hWired‘Our Risk, Their Reward’: Zohran Mamdani on AI and Data Centers18hMIT Technology ReviewHow “reverse game theory” could keep AI agents in check18hArs TechnicaClaude’s new rules against game cloning won’t stop the AI copying boom1dMIT Technology ReviewRaising the quality of insights in agentic analytics1dWiredWhen You Should—and Shouldn’t—Use AI to Search the Web1dWiredHundreds of Mundane AI Apps Are Hiding Sexually Abusive Deepfake Features1dMIT Technology ReviewThe agentic web is here1dWiredChatbots Are Leading Hikers Astray, With Dangerous Consequences2dThe VergeThe biggest modding site in PC gaming is taking a stance on AI4hBBC NewsGraduates are managing AI - but are they learning much else?5hThe VergePlease stop asking me when the AI bubble will pop5hTechCrunchAnthropic is working on a system to prove AI models’ identity6hTechCrunchTed Schlein’s Ballistic Ventures seeks $490M for new funds amid AI-driven cybersecurity boom7hThe VergeApple starts giving Siri AI access to third-party apps9hTechCrunchAI inference startup Fal doubles its valuation to $16B as demand surges9hTechCrunchNetflix’s new AI Discover tab helps you find something to watch based on your mood10hThe VergeOpenAI says ChatGPT is catching up to Google in search11hThe VergeNetflix will show you themed recommendations through a new ‘AI Discover’ tab11hArs TechnicaGoogle will let schools skip AI tools, but they have to give up search, too11hTechCrunchAnthropic is still grappling with the economics of compute12hWiredHere’s Why You Shouldn’t Use AI to Vote14hWired‘Our Risk, Their Reward’: Zohran Mamdani on AI and Data Centers18hMIT Technology ReviewHow “reverse game theory” could keep AI agents in check18hArs TechnicaClaude’s new rules against game cloning won’t stop the AI copying boom1dMIT Technology ReviewRaising the quality of insights in agentic analytics1dWiredWhen You Should—and Shouldn’t—Use AI to Search the Web1dWiredHundreds of Mundane AI Apps Are Hiding Sexually Abusive Deepfake Features1dMIT Technology ReviewThe agentic web is here1dWiredChatbots Are Leading Hikers Astray, With Dangerous Consequences2d