Short Memory — Settings
Configure project name, session TTL, and message limits for short memory.
The Settings tab establishes the foundation for your short memory project. These controls determine how conversation context is stored, when it expires, and how much history your agent can recall.
Basic Information
Project Name
Identifies your short memory project within your account. Must be unique across all short memory projects.
Purpose: Used for dashboard organization and API key association. Choose a name that reflects the agent or application using this memory.
Enable Project
Toggle to activate or deactivate the project. When disabled, all memory operations return an error — no messages can be stored or retrieved.
Purpose: Use this to pause memory operations during maintenance, debugging, or when temporarily disabling an agent.
Session Settings
Session TTL (Time to Live)
Controls how long messages persist after the last activity in a session. Range: 1 minute to 24 hours.
How it affects your pipeline:
- TTL resets on each new message — active conversations stay alive
- When TTL expires, all messages in that session are purged from Redis
- Shorter TTL = less storage cost but agent loses context faster
- Longer TTL = more context available but higher storage usage
Recommendations by use case:
| Use Case | Recommended TTL |
|---|---|
| Live chat support | 30 minutes — 2 hours |
| Task completion workflows | 1 — 4 hours |
| Research sessions | 4 — 8 hours |
| All-day assistants | 12 — 24 hours |
Tip: TTL is measured from the last activity, not from session creation. An active conversation will not expire mid-use.
Maximum Messages
Hard cap on the number of messages stored per session. Range: 5 to 500 messages.
How it affects your pipeline:
- When the limit is reached, the oldest messages are removed to make room for new ones
- Directly controls the context window size your agent receives
- Higher limits provide more conversation history but increase token consumption when retrieving context
- Lower limits keep context focused on recent exchanges
Recommendations by use case:
| Use Case | Recommended Max |
|---|---|
| Simple Q&A | 10 — 20 messages |
| Multi-turn chat | 50 — 100 messages |
| Complex workflows | 100 — 200 messages |
| Full conversation archive | 300 — 500 messages |
Token budget: Retrieving all messages adds to your LLM context window. Balance message count with your model's token limit to avoid truncation or excessive costs.
Next Steps
When your settings are configured, proceed to API Setup to generate your API key and review available endpoints.