# Xynthera AI — JSON invoice, OpenAPI and event checks

> Signed deterministic checks for supplied JSON invoices, OpenAPI API contracts and JSON events.

- name: site.chatgpt.toasty-seal-8877.xynthera-ai/xynthera-ai-prepaid
- version: 2026.10.14
- connection class: R1 (autonomous)
- trust: 83/100 flags: no-repository
- quality: 73/100 (new)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/site.chatgpt.toasty-seal-8877.xynthera-ai%2Fxynthera-ai-prepaid
- tools: https://api.protogrid.dev/v1/servers/site.chatgpt.toasty-seal-8877.xynthera-ai%2Fxynthera-ai-prepaid/tools

**An agent can connect on its own if it already holds `XYNTHERA_AI_PREPAID_TOKEN`.** Remote endpoint behind a static API key or header. The registry never sees the value; the agent fills the placeholder.

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "xynthera-ai-prepaid": {
      "type": "http",
      "url": "https://pfxxzsnboehxsefyajml.supabase.co/functions/v1/xynthera-ai-payg-mcp/mcp",
      "headers": {
        "Authorization": "Bearer ${XYNTHERA_AI_PREPAID_TOKEN}"
      }
    }
  }
}
```

Placeholders to fill: `${XYNTHERA_AI_PREPAID_TOKEN}`.

## Trust

- hygiene: 80
- liveness: 100
- freshness: 100
- provenance: 55
- drivers: +dns-namespace +website +reachable +uptime-100% +updated-5d-ago -no-repository
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 73/100 (new)

### protocol: 75 (weight 25)

- warn `protocol.modern`: newest supported version is 2025-11-25; 2026-07-28 not supported, older versions are deprecated until 2027-07-28
- n/a `protocol.stateless`: only applies to 2026-07-28 servers
- pass `protocol.transport`: streamable HTTP
- n/a `protocol.list_ttl`: only applies to 2026-07-28 servers

### authorization: n/a (weight 15)

- n/a `auth.prm`, `auth.as_metadata`, `auth.cimd`: no remote uses OAuth
- n/a `auth.secret_in_url`: no templated URL

### tool hygiene: 71 (weight 30)

- pass `tools.descriptions`: every tool has a description
- pass `tools.description_length`: descriptions are concise
- pass `tools.schemas`: every tool has an object input schema
- fail `tools.annotations`: no tool declares readOnlyHint or destructiveHint, so clients cannot tell safe calls from risky ones
- fail `tools.directory_hints`: no tool declares all four of readOnlyHint, destructiveHint, idempotentHint and openWorldHint; missing: reconcile_invoice (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); check_contract (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); effect_guard (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); …
- pass `tools.token_cost`: about 2,769 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump

### stability: n/a (weight 15)

- n/a `stability.changes`, `stability.rug_pull`: tool history recorded for 5 of the 7 days needed

### dependencies: n/a (weight 15)

- n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: no npm or PyPI package

- tools: 6, about 2,769 tokens to load them all
- tool set hash: 35226634bb322728c34fe52c34b61265
- tools last changed: 2026-10-15T00:17:35.446Z
- badge: [![protogrid quality](https://protogrid.dev/badge/site.chatgpt.toasty-seal-8877.xynthera-ai/xynthera-ai-prepaid.svg)](https://protogrid.dev/servers/site.chatgpt.toasty-seal-8877.xynthera-ai/xynthera-ai-prepaid)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

- 2026-10-15 `spend_truth` added
  - Create/read a resource budget and reserve or reconcile an agent action against it. A successful new reserve costs 290 micro-EUR excluding tax; exact reserve replays and create_budget/get_budget/get_reservation/begin/settle/release/mark_unknown/reconcile are free. Reservation is the only billable operation and performs no provider action. Each budget accepts at most 128 reservations; a 129th new logical reservation is rejected before billing. Before settle, register the tenant-scoped ES256 outcome verifier described by the hosted OpenAPI. Reuse logical_identity and the exact body after an unknown outcome; keep capabilities private and use the stable provider idempotency key returned by begin.
- 2026-10-15 `effect_guard` added
  - Guard one logical external effect. The prepare operation costs 390 micro-EUR excluding tax only when a new logical decision is successfully created; exact prepare replays and get/claim/settle/reconcile/cancel are free. This service grants one opaque execution authority but never performs the external action. Before settle, register the tenant-scoped ES256 outcome verifier described by the hosted OpenAPI. Reuse the same logical_identity and exact body after an unknown transport outcome; changed content conflicts. Keep claim capabilities private and never retry with a new identity merely because a response was lost.
- 2026-10-14 `check_contract` changed (description)
  - before: Paid EUR 0.010 excluding tax per new signed contract consistency check; checks supplied declarations only. Identical request_id and body avoid a second debit. Report reconstruction is available for 30 days; save the original report. After that the request may be refused, never automatically retried with a new ID.
  - after: Paid EUR 0.010 excluding tax per new signed consistency report for a supplied OpenAPI JSON API contract only, with optional supplied offer/observation comparisons. No legal contract analysis, runtime endpoint testing, security audit or full OpenAPI standard certification. Identical request_id and body avoid a second debit. Report reconstruction is available for 30 days; save the original report. After that the request may be refused, never automatically retried with a new ID.
- 2026-10-14 `get_budget_status` changed (description) **meaning changed**
  - before: Read paid prepaid balance and state; free. The initial purchase amount is the hard server-side total spending limit. No automatic recharge.
  - after: Read authenticated prepaid balance and state; free, with no debit. Before funding, the buyer manually purchases a budget at https://xynthera-ai.toasty-seal-8877.chatgpt.site/payg.html. Requires a private budget code in Authorization: Bearer; no OAuth. The initial purchase amount is the hard server-side total spending limit. No automatic recharge.

Full history: https://api.protogrid.dev/v1/servers/site.chatgpt.toasty-seal-8877.xynthera-ai%2Fxynthera-ai-prepaid/changes

## Remotes

- https://pfxxzsnboehxsefyajml.supabase.co/functions/v1/xynthera-ai-payg-mcp/mcp (streamable-http, auth api_key ${XYNTHERA_AI_PREPAID_TOKEN}, reachable true, uptime30d 1)

## Packages

_none_

## Tools (6)

- `check_contract`: Paid EUR 0.010 excluding tax per new signed consistency report for a supplied OpenAPI JSON API contract only, with optional supplied offer/observation comparisons. No legal contract analysis, runtime endpoint testing, security audit or full OpenAPI standard certification. Identical request_id and body avoid a second debit. Report reconstruction is available for 30 days; save the original report. After that the request may be refused, never automatically retried with a new ID.
- `check_events`: Paid EUR 0.002 excluding tax per signed event batch of up to 200 supplied events, no actual delivery attestation. Identical request_id and body avoid a second debit. Report reconstruction lasts 30 days; retain the returned report. No automatic retries.
- `effect_guard`: Guard one logical external effect. The prepare operation costs 390 micro-EUR excluding tax only when a new logical decision is successfully created; exact prepare replays and get/claim/settle/reconcile/cancel are free. This service grants one opaque execution authority but never performs the external action. Before settle, register the tenant-scoped ES256 outcome verifier described by the hosted OpenAPI. Reuse the same logical_identity and exact body after an unknown transport outcome; changed content conflicts. Keep claim capabilities private and never retry with a new identity merely because a response was lost.
- `get_budget_status`: Read authenticated prepaid balance and state; free, with no debit. Before funding, the buyer manually purchases a budget at https://xynthera-ai.toasty-seal-8877.chatgpt.site/payg.html. Requires a private budget code in Authorization: Bearer; no OAuth. The initial purchase amount is the hard server-side total spending limit. No automatic recharge.
- `reconcile_invoice`: Paid EUR 0.005 excluding tax per new, successful signed invoice consistency check. Supply the entire documented JSON input and a stable unique request_id. Analysis of supplied data only, no bank validation. Exact same-body retries avoid a second debit; report reconstruction lasts 30 days, so retain reports.
- `spend_truth`: Create/read a resource budget and reserve or reconcile an agent action against it. A successful new reserve costs 290 micro-EUR excluding tax; exact reserve replays and create_budget/get_budget/get_reservation/begin/settle/release/mark_unknown/reconcile are free. Reservation is the only billable operation and performs no provider action. Each budget accepts at most 128 reservations; a 129th new logical reservation is rejected before billing. Before settle, register the tenant-scoped ES256 outcome verifier described by the hosted OpenAPI. Reuse logical_identity and the exact body after an unknown outcome; keep capabilities private and use the stable provider idempotency key returned by begin.