Memro Python SDK
Verified memory infrastructure for AI agents.
Install
# PyPI (once published via CI):
pip install memro
# Install from source:
pip install .
Quick Example
from memro import Agent
agent = Agent.create() # Generates sovereign Ed25519 identity
mem = agent.remember("The billing service was rewritten in Go")
results = agent.recall("billing service language")
print(results[0].content)
Connect
from memro import Agent
agent = Agent.from_env() # MEMRO_API_KEY (and optional MEMRO_BASE_URL)
agent = Agent.from_api_key("memro_sk_...") # explicit key, Memro Cloud
agent = Agent.connect("memro_sk_...", base_url="http://localhost:8081") # self-hosted
Self-owned identity (the private key is generated here and never sent):
agent = Agent.create() # registers the public key
agent.save("identity.json") # keep secret; it can't be recovered
agent = Agent.from_file("identity.json")
Agent.from_env() also accepts MEMRO_AGENT_ID + MEMRO_PRIVATE_KEY.
Memories
m = agent.remember(
"The billing service was rewritten in Go",
type="semantic", # episodic | semantic | procedural | profile | None (auto)
visibility="private",
event_date="2026-03-01T00:00:00Z", # optional
session_id=None, # optional UUID
metadata={"source": "standup"},
)
agent.get_recent(limit=20, type="profile") # list[Memory], newest first
agent.delete_memory(m.id) # True, or False if it didn't exist
agent.export() # ExportData(total, memories, ...)
Search
for r in agent.recall("what language is billing in?", limit=5, mode="hybrid", min_score=0.3):
print(r.score, r.content) # SearchResult
agent.recall_temporal("deploy incidents", after_date="2026-09-01T00:00:00Z", temporal_weight=0.6)
mode is hybrid (default), semantic or keyword.
Memory engine
ans = agent.reason("why did 500s spike after the billing release?")
print(ans.answer, ans.confidence, ans.based_on)
agent.evolve() # EvolutionResult
agent.resolve_conflict(a_id, b_id) # {"Resolved": id} or {"Ambiguous": [...]}
agent.memory_chain(m.id) # MemoryChain
agent.timeline(m.id) # MemoryTimeline
agent.stats() # TemporalStats
agent.usage() # dict
Files and webhooks
up = agent.upload_file("notes.pdf") # stores the file and a memory of its text
up = agent.upload_file(("notes.txt", b"...", "text/plain"), remember=False)
agent.upload_image("whiteboard.png") # Memory with a description
agent.upload_document("handbook.docx") # PDF/Word/PowerPoint/Excel/text -> list[Memory]
agent.add_webhook("https://example.com/memro-hook")
agent.webhooks()
agent.remove_webhook("https://example.com/memro-hook")
Errors and retries
from memro import MemroError, RateLimitError
try:
agent.remember("...")
except RateLimitError:
... # 429 after the automatic retries
except MemroError as e:
print(e.status_code, e.message)
MemroClient(timeout=30, max_retries=2): rate-limited requests are retried
(honouring Retry-After); gateway errors are retried only for GET and
DELETE, so a memory is never stored twice.
LangChain
from memro.integrations.langchain import MemroMemory # pip install "memro[langchain]"
memory = MemroMemory(agent=agent)
Low-level client
agent.client is a MemroClient with one method per endpoint
(search, search_temporal, reason, upload_file, …) and
request(method, path, params=, json_body=, files=) for anything else.
Requests are signed with v2 signatures
when there is no API key.
See the API reference for response fields.
Repository: github.com/memrohq/memro · Documentation: https://memro.co License: Source available under PolyForm Noncommercial 1.0.0. Free for non-commercial use; commercial use through Memro Cloud or a commercial license.
Metadata
Release files for memro 0.2.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| memro-0.2.0.tar.gz | 23.8 kB | Without provenance | Details for memro-0.2.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| memro-0.2.0-py3-none-any.whl | Python 3 | none | any | Details for memro-0.2.0-py3-none-any.whl |
Total release size: 43.4 kB
Release files / memro-0.2.0.tar.gz
| Download URL | memro-0.2.0.tar.gz |
|---|---|
| Size | 23.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
a9a2bb789270e86ba7185cb3e4ba0a710e5cee274ac47a80bfedcc5feab9ca9a
|
|
BLAKE2b-256 checksum How to use checksums |
3bc93e9c38b78b001aee5132da0076764c78f6be51ceecece7b17d4981af7855
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.15.1
|
Release files / memro-0.2.0-py3-none-any.whl
| Download URL | memro-0.2.0-py3-none-any.whl |
|---|---|
| Size | 19.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
830ae06054688ba75f26205ccdcdd7a2777157cbc344359d077be20c76f32989
|
|
BLAKE2b-256 checksum How to use checksums |
4bdf1ab4a9cf8ee8f0b3e57469e995e7effd99f4dd36eda7bde86d19464b37f5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.15.1
|
Security for memro 0.2.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.