> ## Documentation Index
> Fetch the complete documentation index at: https://praison.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Bounded Learning
> Keep continuous-learning stores healthy by capping size and expiring stale entries.
Keep continuous-learning stores healthy by capping size and expiring stale entries — automatically, without any LLM judgement.
```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from praisonaiagents import Agent, LearnConfig
agent = Agent(
name="Assistant",
instructions="Adapts without unbounded memory growth",
learn=LearnConfig(persona=True, max_entries=1000, retention_days=90),
)
agent.start("Remember I prefer bullet points")
```
The user keeps teaching the agent; old or excess learnings archive automatically so search stays fast.
```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
graph LR
subgraph "Bounded Learning"
A[💡 New learning] --> B[💾 Store entry]
B --> C[🔁 Used in search]
C --> D[📈 use_count++]
D --> E{🚦 Over cap?}
E -->|No| F[Keep]
E -->|Yes| G[📦 Archive sidecar]
G --> H[Active store stays small]
end
classDef input fill:#6366F1,stroke:#7C90A0,color:#fff
classDef proc fill:#F59E0B,stroke:#7C90A0,color:#fff
classDef ok fill:#10B981,stroke:#7C90A0,color:#fff
classDef archive fill:#8B0000,stroke:#7C90A0,color:#fff
class A,B input
class C,D proc
class E,F,H ok
class G archive
```
## How It Works
```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
sequenceDiagram
participant User
participant Agent
participant BoundedLearning
User->>Agent: Request
Agent->>BoundedLearning: Process
BoundedLearning-->>Agent: Result
Agent-->>User: Response
```
## Quick Start
Set `max_entries` to limit how many learnings each store retains:
```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from praisonaiagents import Agent, LearnConfig
agent = Agent(
name="Assistant",
instructions="Adapts to user preferences",
learn=LearnConfig(
persona=True,
max_entries=1000, # Keep at most 1000 entries per store
retention_days=90, # Mark entries older than 90 days as stale
),
)
agent.start("Help me plan my day")
```
Call `prune()` on the learn manager for one-off cleanup:
```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from praisonaiagents import Agent, LearnConfig
agent = Agent(
learn=LearnConfig(max_entries=1000, retention_days=90)
)
# Returns evicted count per store
evicted = agent.memory.learn().prune()
print(f"Pruned: {evicted}")
# e.g. {"persona": 12, "insights": 3}
```
***
## How It Works
```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
sequenceDiagram
participant User
participant Agent
participant Feature as Bounded Learning
User->>Agent: Request
Agent->>Feature: Process request
Feature-->>Agent: Result
Agent-->>User: Response
```
Usage telemetry is observed automatically — every `search()` and `list_all()` call increments `use_count` and updates `last_used` on returned entries. No extra configuration needed.
| What triggers pruning | When |
| --------------------- | --------------------------------------------------------------------------- |
| On write | Every time an entry is added when `max_entries > 0` and the cap is exceeded |
| On `prune()` call | Manually, for one-off cleanup of a long-running agent |
| Time-based | Entries older than `retention_days` with no recent use are eligible |
**Eviction scoring:** entries are ranked by `use_count` ascending, then `created_at` ascending. The least-used, oldest entries are evicted first — the most valuable, recently accessed learnings always survive.
**Archival:** evicted entries are appended to a per-store `.archive.json` sidecar (e.g., `persona.archive.json`) in the same directory as the active store file. The archive is append-only JSON — nothing is permanently hard-deleted.
***
## Configuration Options
Full reference on `max_entries` and `retention_days` is in the [LearnConfig reference](/docs/configuration/learn-config).
```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from praisonaiagents import LearnConfig
LearnConfig(
max_entries=0, # 0 = unbounded (default)
retention_days=0, # 0 = never expires (default)
)
```
Both fields default to `0` (unbounded). Existing agents are completely unaffected unless you opt in by setting a non-zero value.
***
## Best Practices
If your agent runs for short sessions or handles a limited number of users, the defaults (`max_entries=0`, `retention_days=0`) are fine. Stores only grow with real interactions and stay small naturally.
Agents that run 24/7 and talk to many users can accumulate thousands of entries over time. Set `max_entries=1000` to keep store files small and keep search fast. Pruning runs automatically on every write — no cron job needed.
```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
learn=LearnConfig(
persona=True,
insights=True,
max_entries=1000,
)
```
For data minimization requirements (GDPR, SOC 2, etc.), set `retention_days` to make stale entries eligible for pruning. Pair with `max_entries` so pruning triggers automatically on write rather than requiring a manual call.
```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
learn=LearnConfig(
persona=True,
max_entries=500,
retention_days=30, # 30-day data retention policy
)
```
Evicted entries are never permanently deleted. The `.archive.json` sidecar (same directory as the store file) is an append-only JSON array. Open it to audit or restore evicted entries.
Archive file location example:
```
~/.praison/learn/private/default/persona.archive.json
```
***
## Related
Full configuration options for continuous learning
Save and restore agent state across runs