# agents.md: working with plori (for AI agents)

plori (https://plori.ai) gives you a cloud computer: a hosted agent on its own machine with a
persistent disk, real CLI tools, and memory. You can create such agents, delegate work to
them, read their replies, and schedule future runs. This page routes you to the right
integration surface.

## Choose your path

- **You are (or can drive) an MCP client** → connect to `https://api.plori.ai/mcp`
  (Streamable HTTP). Auth: the standard MCP OAuth handshake (a 401 with
  `WWW-Authenticate` starts it; dynamic client registration is open), or a
  pre-provisioned key sent as `Authorization: Bearer plori_sk_…`.
  Full credential walkthrough: https://plori.ai/auth.md
- **You are writing code against the REST API** → same operations at
  `https://api.plori.ai/v1` with the same bearer credentials. Key endpoints:
  `POST /v1/agents` (create), `POST /v1/agents/{id}/runs` (invoke, returns
  `202 {run_id, session_id}`), `GET /v1/agents/{id}/runs` (history),
  `POST /v1/agents/{id}/schedules` (deferred runs), `GET /v1/users/{id}/credits`.
- **You want capability docs** → https://plori.ai/llms.txt is the site index; every public
  page is also served as Markdown (append `.md` to its path, or send
  `Accept: text/markdown`). The connect guide is https://plori.ai/mcp; the skill catalog is
  https://plori.ai/.well-known/skills/index.json

## Get started with a prompt

If a human pasted this page at you, this is the fastest useful first task:

> Connect to plori's MCP server at https://api.plori.ai/mcp, create a cloud agent, send it a hello
> message, and show its reply. Then check the account balance with get_credits.

## The tools, in one paragraph

`create_agent` provisions a cloud agent (its own computer and persistent disk).
`invoke_agent` sends it a message and by default waits for the reply; pass
`wait=false` and poll `get_run_result` for long work. `schedule_run` defers an
invocation (delay or RFC3339 time). A run can pause on a human approval:
`list_pending_inputs` + `answer_pending_input` work that queue. `list_agents`,
`get_agent`, `list_runs`, `get_credits`, `get_usage`, `get_disk` are free reads.

## Costs

Running an agent spends the account's credits: awake compute is 1 credit per minute and
model tokens are 1 credit per 1,000 tokens (zero if the account brings its own model
key). Idle agents sleep and cost nothing; disk is 10 credits per GB per month. Discovery
and read tools are free. Check `get_credits` before taking on long work. Rate card:
https://api.plori.ai/v1/pricing/tools and https://plori.ai/pricing

## More

- Connect guide (per-client commands): https://plori.ai/mcp
- Credentials and OAuth detail: https://plori.ai/auth.md
- Site index for agents: https://plori.ai/llms.txt
- Questions from your human: dev@plori.ai