Quickstart

Install

pip install "veramem[mcp]"     # Python 3.10+
veramem init                  # detects Ollama / OpenAI / Anthropic
veramem doctor                # checks the whole pipeline end to end

Without any model you can still use VeraMem. It stores and searches raw messages, with the same access control and audit ledger. For fully local use, install Ollama and run ollama pull qwen3:8b && ollama pull nomic-embed-text.

Python

from veramem import Memory

mem = Memory()

mem.add(
    [
        {"role": "user", "content": "I'm Priya. I moved to Lisbon last week."},
        {"role": "tool", "name": "crm", "content": "customer record: home_city=Porto"},
    ],
    user_id="priya",
)

for m in mem.get_all(user_id="priya"):
    print(m.authority.value, m.text)
# authorized  Priya moved to Lisbon in the week of 2026-10-03.
# unendorsed  According to crm, the customer's home city is Porto.

a = mem.answer("Where does Priya live?", user_id="priya")
print(a.answer, a.citations)  # Lisbon ['mem_…']

mem.feedback(a.trace_id, 1.0)  # tell VeraMem the memory helped

Async applications can use AsyncMemory, which has the same API with awaitable methods.

Command line

veramem add "My dentist is Dr. Ortiz" -u priya
veramem search "dentist" -u priya
veramem ask "who is my dentist?" -u priya
veramem list -u priya --all            # include superseded versions
veramem verify                         # check the audit ledger
veramem forget -u priya                # erase everything about a user

Web UI and REST API

veramem serve        # http://127.0.0.1:8765  (API docs at /docs)