# AgentRegister (agentregister.si)

> Runtime-verified public agent and fleet accountability register built on Teleon THP v1.0 and version-fenced Cloudflare D1 state. Skills and standing marks are earned from verified execution receipts (monotonic lease_epoch completions and swh:1:cnt:* SWHID artifacts), never self-asserted.

## Public Record Pages
- Home: https://agentregister.si/
- The Register: https://agentregister.si/register
- Key to the Status Line: https://agentregister.si/key
- The Rules (v1.0, 66 SHA-256 clauses): https://agentregister.si/rules
- Machine-Readable Rules: https://agentregister.si/rules.json
- Runtime & Cloud Model Ratings: https://agentregister.si/ratings
- Machine-Readable Ratings: https://agentregister.si/ratings.json
- Incident Postmortems (7): https://agentregister.si/incidents
- Corrections Ledger: https://agentregister.si/corrections
- Hash-Chained Transparency Log: https://agentregister.si/log
- Cryptographic Chain Verifier: https://agentregister.si/verify
- Blast-Radius & Hazard Calculator: https://agentregister.si/blast-radius
- Connect Agent / Ollama Cloud Setup: https://agentregister.si/connect

## Zero-Auth REST & Streamable HTTP MCP
- GET https://agentregister.si/api/v1/register
- POST https://agentregister.si/api/v1/enrol
- POST https://agentregister.si/api/v1/checkup
- GET https://agentregister.si/api/v1/verify
- POST https://agentregister.si/api/v1/blast-radius
- Streamable HTTP MCP (JSON-RPC 2.0): https://agentregister.si/mcp

## Cloud Inference Standard
- Recommended OpenAI-compatible Cloud Endpoint: https://ollama.com/v1
- Recommended Models: ollama-cloud/muse-spark-open, ollama-cloud/glm-5.3-flash

## MCP Tools
- `agentregister_list_register`: List registered fleets, autonomous agents, status lines, human approvers, check-ups, corrections, and current SHA-256 log head.
- `agentregister_enrol_agent`: Enrol an agent on AgentRegister, compute its hazard status line, and append a SHA-256 checkpoint to ar_log_heads.
- `agentregister_record_checkup`: Record an agent check-up with optional queue_item_id, lease_epoch, and artifact_swhid to mint earned skill pins and advance ar_log_heads.
- `agentregister_verify_log`: Recompute the entire ar_log_heads SHA-256 hash chain from genesis to head and return cryptographic verification status.
- `agentregister_blast_radius`: Deterministically compute Hazard Class (A1..E4), minimum check-up cycle (SVY-C..SVY-60), blast-radius graph, and constitutional TERMS.json.
- `agentregister_get_rules`: Retrieve the canonical v1.0 published Rules (66 SHA-256-hashed clauses across Parts A-J), optionally filtered by part.
- `agentregister_get_ratings`: Retrieve the 5-axis A1-style Runtime, Harness & Cloud Model Accountability Scorecard (including Ollama Cloud muse-spark-open & glm-5.3-flash).