Framework-agnostic agent persistence platform. Local-first, portable AgentSoul memory.
Project description
AgentSoul — Framework-Agnostic Agent Persistence Platform
The memory backbone for production AI agents. Works with any LLM framework.
What is AgentSoul?
AgentSoul is a standalone, monetizable infrastructure product that solves the hardest problem in agent development: persistent, reliable, portable agent memory.
Unlike chatbots that forget after each session, AI agents in production need:
- Long-term memory of customers, interactions, and patterns
- Memory portability (dev Mac → customer RPi5 → fleet of devices)
- Zero drift (consolidation + decay prevents hallucination)
- Framework flexibility (works with Hermes, LangGraph, CrewAI, AutoGen, etc.)
- Compliance-ready (audit trails, GDPR delete, encryption)
AgentSoul provides all of this in <10 lines of Python code.
Core Features
1. Universal Memory Interface (SDK)
One Python class, three backends: SQLite (standalone), REST (cloud), PocketBase (local server).
from agentsoul import AgentSoul
# Initialize once, works everywhere
soul = AgentSoul.from_rest(
url="http://localhost:5001",
agent_id="clerk_001",
token="your_api_token"
)
# Store memory
soul.remember("customer", "cust_001", {
"name": "Alice",
"payment_style": "reliable",
"contact": "sms"
})
# Retrieve with automatic consolidation
context = soul.recall("customer", "cust_001", consolidate=True)
# Inject into agent
system_prompt = soul.get_system_context("cust_001")
2. Importance-Based Decay (Not just time)
Memories decay based on business signals, not just age:
- Recent interactions get higher weight
- Frequently-mentioned facts persist longer
- High-value customers = higher memory weight
- Explicit feedback signals boost importance
Result: Agents stay focused on what matters.
3. Soul Portability (Export/Import)
Move a complete agent soul with one command:
# On Mac (dev)
artifact = soul.export_soul(passphrase="secure_123")
# Save artifact for transfer to RPi5
# On RPi5 (production)
soul_prod.import_soul(artifact, "secure_123")
# Complete state instantly available
4. System Prompt Injection
Pre-loaded context before every agent interaction:
You are serving customer_001.
CONSOLIDATED MEMORY:
- Reliability: highly reliable (95%)
- Prefers: SMS, morning
- Service types: HVAC repair, filter replacement
- Notes: Always pays on time
Respond naturally and helpfully.
5. Production-Ready Packaging
- REST API with Bearer token auth
- Docker support for cloud/Kubernetes
- RPi5 one-liner deployment
- PocketBase integration (local SQLite server)
- Audit trails for compliance
Comparison: AgentSoul vs Alternatives
| Feature | AgentSoul | Mem0 | Zep | Letta |
|---|---|---|---|---|
| Framework Agnostic | ✓ Any LLM | ✓ | ✓ | ✗ (Letta-specific) |
| Soul Portability | ✓ Export/import | ✗ | ✗ | ✗ |
| Local-First | ✓ SQLite + REST | ✗ (SaaS) | ✗ (SaaS) | ✓ |
| Importance Decay | ✓ Business-signals | ✗ (Time-only) | ✗ (Time-only) | ? |
| Encryption | ✓ AES-256-GCM | ✗ | ✗ | ? |
| GDPR Ready | ✓ Delete + audit | ✗ | ✗ | ? |
| Standalone License | ✓ MIT/Commercial | ✗ (SaaS) | ✗ (SaaS) | ✓ |
| Python SDK | ✓ <10 lines | ? | ✗ | ✓ |
| RPi5 Deploy | ✓ One-liner | ✗ | ✗ | ✗ |
| Cost | $0-999/yr | $9-99/mo | $0-299/mo | Free |
AgentSoul wins on: Portability + Local-first + Framework flexibility
Quick Start
1. Install AgentSoul
git clone https://github.com/agentsoul/agentsoul
cd agentsoul
pip install -e .
2. Initialize Database
python -m agentsoul.persistence.schema
# Creates SQLite tables
3. Start REST API (optional)
python -m agentsoul.adapters.rest_api
# Listening on http://127.0.0.1:5001
4. Use in Your Agent
from agentsoul import AgentSoul
soul = AgentSoul.from_sqlite(
db_path="/path/to/db.db",
agent_id="my_agent"
)
# Store memory
soul.remember("entity", "user_123", {"age": 30, "role": "buyer"})
# Recall with consolidation
memory = soul.recall("entity", "user_123")
# Get system prompt
prompt = soul.get_system_context("user_123")
Licensing & Pricing
Open Source (MIT)
- Free for development and educational use
- Local SQLite backend only
- Full source code
Commercial (Annual)
| Tier | Price | Agents | Support |
|---|---|---|---|
| Startup | $299/yr | 10 | |
| Professional | $999/yr | 100 | Priority |
| Enterprise | Custom | ∞ | Dedicated SLA |
Use the open-source version for free. Pay only if you deploy commercially.
Use Cases
1. Mobile Service Techs (HVAC, Plumbing, Electrical)
Portable clerk agent on RPi5, carried to every job site.
2. Virtual Assistant Services
Agents that remember customer preferences across devices.
3. Conversational AI (SaaS)
Long-term memory for chatbots that scale.
4. Enterprise Agent Fleet
Multiple agents sharing memory via central PocketBase.
5. AI Research
Persistent memory for agent experiments without drift.
Architecture
┌─────────────────────────────────────────────┐
│ Agent Framework │
│ (Hermes, LangGraph, CrewAI, AutoGen) │
└──────────────────┬──────────────────────────┘
│
▼
┌──────────────────────┐
│ AgentSoul SDK │
│ (Python interface) │
└──────────────┬───────┘
│
┌────────────┼────────────┐
▼ ▼ ▼
┌─────────┐ ┌──────────┐ ┌────────┐
│ SQLite │ │ REST API │ │ MCP │
│ Local │ │ (HTTP) │ │Server │
└────┬────┘ └────┬─────┘ └────┬───┘
│ │ │
└────────────┼─────────────┘
▼
┌────────────────────────────┐
│ PocketBase Data Store │
│ (agent_soul_memories, │
│ interactions, audit) │
└────────────────────────────┘
Security & Compliance
- Encryption: AES-256-GCM (NIST-grade)
- Key Derivation: PBKDF2-HMAC-SHA256 (480k iterations)
- Audit Trails: All actions logged
- GDPR: DELETE endpoint for memory removal
- Authentication: Bearer tokens (pluggable)
Roadmap
- Vector search (semantic similarity)
- Multi-agent collaboration
- Graph memory (relationship mapping)
- LLM-powered summarization
- Web UI dashboard
- Kubernetes operator
- OpenAI Assistant integration
Contributing
Contributions welcome! Submit PRs to GitHub.
Support
- Docs: https://agentsoul.docs
- GitHub Issues: https://github.com/agentsoul/agentsoul/issues
- Email: support@agentsoul.io
- Discord: Join community
AgentSoul: The memory backbone your agents deserve.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file agentsoul-0.1.0.tar.gz.
File metadata
- Download URL: agentsoul-0.1.0.tar.gz
- Upload date:
- Size: 17.1 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.9.6
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
d208253dcd9a5f66a2e70a7f64807d8609a0b9f6cf83717e1e216a5ec1fb9545
|
|
| MD5 |
69874ce5bcdca655180b3060fba313a0
|
|
| BLAKE2b-256 |
1f8e47c8cb3ebf50d614538b05126fa75fd9e4b8122ac4cd7a1f711f92167d7d
|
File details
Details for the file agentsoul-0.1.0-py3-none-any.whl.
File metadata
- Download URL: agentsoul-0.1.0-py3-none-any.whl
- Upload date:
- Size: 17.1 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.9.6
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
ba0e8cc5cefd86379938c014ac8b6aa00f07d8dd0faf619e45e61dde16dca37e
|
|
| MD5 |
3414330b802d9ca9df9c533009ce1a8e
|
|
| BLAKE2b-256 |
31e1ae010e51debc474c83b6d1aeb53096cb9b11398d25e3d9d7398a03250f21
|