Python SDK

pip install octopoda

from octopoda import Octopoda
 
client = Octopoda(api_key="sk-octopoda-YOUR_KEY")
agent = client.agent("my-agent")
 
# Store a memory
agent.write("user:name", "Alice")
 
# Read it back
name = agent.read("user:name")
print(name)  # Alice
 
# Search by meaning
results = agent.search("what is the user's name?")
print(results)

client = Octopoda(
    api_key="sk-octopoda-...",                    # or set OCTOPODA_API_KEY env var
    base_url="https://api.octopodas.com",         # default
    timeout=30,                                    # seconds
)

# Delete specific memory
agent.forget("old_key")
 
# Delete old memories (preserves critical importance)
agent.forget_stale(max_age_seconds=604800)  # 7 days
 
# Delete by tag
agent.forget_by_tag("temporary")
 
# Find duplicates (preview — changes nothing)
report = agent.consolidate(dry_run=True)
print(f"Found {report['duplicate_groups']} duplicate groups")
 
# Merge duplicates (keeps newest, removes duplicates)
result = agent.consolidate(dry_run=False)
 
# Memory health score
health = agent.memory_health()
print(f"Score: {health['score']}/100")
for issue in health['issues']:
    print(f"  - {issue}")

agent.set_goal("Migrate database", milestones=["Backup", "Migrate", "Validate"])
agent.update_progress(milestone_index=0, note="Backup done")
goal = agent.get_goal()
# {"progress": 0.33, "status": "active"}

agent.send_message("agent_b", "Found a bug", message_type="alert")
messages = agent.read_messages(unread_only=True)
agent.broadcast("Deployment starting")

bundle = agent.export_memories()           # Export as JSON
agent.import_memories(bundle, overwrite=False)  # Import (skip existing)

results = agent.search_filtered(
    query="deployment issues",
    tags=["production"],
    importance="critical",
    max_age_seconds=86400
)

result = agent.process_conversation([
    {"role": "user", "content": "I'm Alice, CTO at TechCorp"},
    {"role": "assistant", "content": "Nice to meet you! How can I help?"},
    {"role": "user", "content": "We need to optimize our ML pipeline"},
])
 
print(f"Stored {result['memories_stored']} memories")
 
# Get relevant context for a query
context = agent.get_context("what does the user want?")
print(context)

Install Octopoda memory for my Python project:
 
1. Run: pip install octopoda
2. Test it with this code:
 
from octopoda import Octopoda
client = Octopoda(api_key="sk-octopoda-MY_KEY_HERE")
agent = client.agent("test-agent")
result = agent.write("test", "hello world")
print("Write:", result)
value = agent.read("test")
print("Read:", value)
 
Tell me if both succeed.