AgentsPodium Hosting
Measured conclusion: Alive, not deep-tested. L2 outcome: not_attempted.
Create, watch, pay for and connect hosted AI agent pods on AgentsPodium. Needs an API key.
| Author | com.agentspodium |
|---|---|
| Access | remote |
| Publisher key required | No |
| Measured state | Alive, not deep-tested |
| L1 liveness | alive |
| L1 latency | 366 ms |
| Last measured | 2026-10-26 |
Add to your agent
Copy this into your MCP client config:
{
"mcpServers": {
"AgentsPodium Hosting": {
"url": "https://mcp.agentspodium.com/mcp",
"transport": "streamable-http"
}
}
}
Measured tools (13)
list_platformsread-onlyList available AgentsPodium engines (platforms) and hosting plans (tiers), with prices and capabilities.
Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes
list_instancesread-onlyList the caller's AgentsPodium instances (agents): id, name, engine, tier, status, endpoint and A2A URL, domain, creation date. Never returns secrets.
Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes
create_instanceCreate (deploy) a new AgentsPodium instance. Returns the instance id and, when enableA2A is true, the a2aUrl and a2aToken the caller should keep. After creating, poll get_instance_health until it is serving, then call set_llm_key to give it a working LLM key.
Parameters:
tier(required) — string — Hosting plan controlling RAM/CPU/disk. See list_platforms for prices.name— string — Display name for the instance.model— string — Default LLM model id for the engine.domain— string — Customer-owned domain to serve the instance from.engine— string — Runtime engine id (e.g. hermes, openclaw, n8n). See list_platforms for what's available.channels— array — Channels to enable (web, telegram, ...).enableA2A— boolean — Turn on the A2A toolset and issue an a2aUrl/a2aToken for this instance.extraSoul— string — Extra system-prompt instructions appended to the persona.personaId— string — Persona/template id from list_platforms; defaults to the general-purpose persona.webhookUrl— string — Public http(s) URL to receive signed pod events (agent.running, agent.stopped, agent.failed, agent.deleted, payment.confirmed, deletion.warning) instead of polling. The signing secret comes back once as webhookSecret.
Hints: read-only not declared · destructive not declared · idempotent not declared · open-world yes
get_instance_healthread-onlyCheck whether an instance is reachable and serving, plus its status and quota state.
Parameters:
id(required) — string — Instance (agent) id, as returned by create_instance or list_instances.
Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes
get_instance_termread-onlyGet the billing term (trial/paid/grace) for an instance, including when it renews or expires.
Parameters:
id(required) — string — Instance (agent) id, as returned by create_instance or list_instances.
Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes
set_llm_keySet (or clear, with key=null) the LLM provider API key an instance uses to answer. Required before a freshly created instance can actually respond.
Parameters:
id(required) — string — Instance (agent) id, as returned by create_instance or list_instances.provider(required) — string — LLM provider id, e.g. openai, anthropic, deepseek.key(required) — string — The provider API key. Stored encrypted, never echoed back.model— string — Model id to use with this provider (e.g. gpt-4o-mini). Optional; the engine defaults if omitted.
Hints: read-only not declared · destructive not declared · idempotent yes · open-world yes
pause_instancePause an instance: stops it and its billing lease. Data is archived, not deleted.
Parameters:
id(required) — string — Instance (agent) id, as returned by create_instance or list_instances.
Hints: read-only not declared · destructive not declared · idempotent yes · open-world yes
resume_instanceResume a previously paused instance, rebuilding it from its archive.
Parameters:
id(required) — string — Instance (agent) id, as returned by create_instance or list_instances.
Hints: read-only not declared · destructive not declared · idempotent yes · open-world yes
rebuild_instancedestructiveRebuild an instance's pod (e.g. after a config change that needs a restart).
Parameters:
id(required) — string — Instance (agent) id, as returned by create_instance or list_instances.
Hints: read-only not declared · destructive yes · idempotent not declared · open-world yes
delete_instancedestructivePermanently delete an instance. This cannot be undone.
Parameters:
id(required) — string — Instance (agent) id, as returned by create_instance or list_instances.
Hints: read-only not declared · destructive yes · idempotent not declared · open-world yes
get_payment_optionsread-onlyGet ways to pay for an instance's plan: card/subscription buy links, crypto payment info, and Telegram Stars info (crypto and Stars are in beta testing), plus the instance's current term.
Parameters:
id(required) — string — Instance (agent) id, as returned by create_instance or list_instances.
Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes
set_peersSet the list of other A2A agents this instance is allowed to call by name.
Parameters:
id(required) — string — Instance (agent) id whose peer list is being set.peers(required) — array — Full replacement list of peer agents this instance may call (up to 20).
Hints: read-only not declared · destructive not declared · idempotent yes · open-world yes
set_webhookRegister a URL that receives signed POSTs when the instance comes up, stops, fails, is deleted, when a payment is confirmed and before deletion for non-payment — instead of polling get_instance_health. Returns the HMAC secret once.
Parameters:
id(required) — string — Instance (agent) id, as returned by create_instance or list_instances.url(required) — string — Public http(s) URL to POST signed pod events to. Private and loopback hosts are refused.
Hints: read-only not declared · destructive not declared · idempotent yes · open-world yes
Measured by MCPYP probing. Untested fields are omitted, never estimated. Listing is always free; MCPYP never proxies calls or stores user tokens.