SPM SPM Docs

Quickstart

Create a memory-backed request in about five minutes.

1. Get your API key

Sign in to the console at app.spmos.ai, then create an API key under API Keys. Keys look like spm_live_… and are shown once. Store the key securely.

2. Add a provider credential

Under Providers, add the LLM account that SPM should use. Select the provider type, enter its endpoint and API key, and choose a default model. Provider secrets are vaulted and never appear in the browser, logs, or receipts after saving. See Bring your own provider.

3. Make a call

OpenAI SDK (Python):

from openai import OpenAI

client = OpenAI(
    base_url="https://api.spmos.ai/v1",
    api_key="spm_live_...",          # your SPM key, not the provider key
)

reply = client.chat.completions.create(
    model="deepseek-chat",           # any model your provider credential serves
    messages=[{"role": "user", "content": "Remember: my deploy window is Friday."}],
)
print(reply.choices[0].message.content)

Anthropic SDK:

import anthropic

client = anthropic.Anthropic(
    base_url="https://api.spmos.ai",
    api_key="spm_live_...",
)

curl:

curl https://api.spmos.ai/v1/chat/completions \
  -H "Authorization: Bearer spm_live_..." \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek-chat","messages":[{"role":"user","content":"Hello"}]}'

Migration requires only the base URL + key. The rest of your agent code does not change.

4. See memory work

Send several turns, then ask about an earlier statement. SPM recalls relevant memory through the evidence gate and adds it to the prompt. On our benchmark harness, this added 791 tokens instead of the complete history.

5. Watch it in the console

Each request produces a signed receipt covering usage, latency, and memory tokens. Sources, audit events, and purge receipts are available under the console's Audit and Sources tabs.

Next steps