Skip to content

Add Redis Agent Memory developer guide and improve index lede#3553

Closed
mich-elle-luna wants to merge 2 commits into
mainfrom
agent-mem-updates
Closed

Add Redis Agent Memory developer guide and improve index lede#3553
mich-elle-luna wants to merge 2 commits into
mainfrom
agent-mem-updates

Conversation

@mich-elle-luna

@mich-elle-luna mich-elle-luna commented Jun 25, 2026

Copy link
Copy Markdown
Collaborator
  • Add developer-guide.md: complete reference covering architecture, quick starts for Redis Cloud and on-premises, memory concepts, lifecycle, store management, authentication, configuration reference, and agent code examples (Python + REST)
  • Update _index.md lede to lead with the "why" — framing the memory infrastructure problem agents face before introducing RAM as the solution
  • Add "Redis Cloud setup and management" section to _index.md with direct links to the operate/rc create-service, use-agent-memory, and view-service pages
  • Add operate/rc cross-reference note at the top of the Redis Cloud quick start section in the developer guide

Note

Low Risk
Documentation-only changes with no runtime or API behavior impact.

Overview
Adds a complete developer guide for Redis Agent Memory and tightens how the section introduces the product.

The new developer-guide.md is a single reference for Redis Cloud and on-premises: architecture, Cloud and Helm quick starts, memory tiers and lifecycle, store management, auth (API keys, iris_ agent keys, OIDC), Helm/env configuration, and Python/httpx plus REST examples.

The Agent Memory index now opens with the “why” (agent memory infrastructure pain) before positioning RAM as the managed solution, and adds a Redis Cloud setup and management block linking to operate/rc create-service, use-agent-memory, and view-service pages. The Cloud quick start in the developer guide points to those same operate docs.

Reviewed by Cursor Bugbot for commit 0e23d2b. Bugbot is set up for automated code reviews on this repo. Configure here.

- Add developer-guide.md: complete reference covering architecture,
  quick starts for Redis Cloud and on-premises, memory concepts,
  lifecycle, store management, authentication, configuration reference,
  and agent code examples (Python + REST)
- Update _index.md lede to lead with the "why" — framing the memory
  infrastructure problem agents face before introducing RAM as the solution
- Add "Redis Cloud setup and management" section to _index.md with
  direct links to the operate/rc create-service, use-agent-memory,
  and view-service pages
- Add operate/rc cross-reference note at the top of the Redis Cloud
  quick start section in the developer guide

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
@mich-elle-luna mich-elle-luna requested a review from a team June 25, 2026 19:17

@cursor cursor Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

Fix All in Cursor

❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.

Reviewed by Cursor Bugbot for commit 64cd86f. Configure here.

Comment thread content/develop/ai/context-engine/agent-memory/developer-guide.md
@jit-ci

jit-ci Bot commented Jun 25, 2026

Copy link
Copy Markdown

🛡️ Jit Security Scan Results

CRITICAL HIGH MEDIUM

✅ No security findings were detected in this PR


Security scan by Jit

@dwdougherty dwdougherty left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Language LGTM.

add_event hardcoded actorId 'user-1' while search filtered by
ownerId 'user-alex', so promotion scoping and retrieval didn't align.
Pass owner_id as actor_id for user events so the stored actorId
matches the ownerId used in memory search.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

@atanas-andonov atanas-andonov left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are some features described which do not exist, I suspect they could be taken from the predecessor of agent memory here: https://github.com/redis/agent-memory-server

The docs should only contain features implemented by this repository: https://github.com/redislabsdev/iris

Comment on lines +156 to +157
- **LLM provider** — select your embedding provider (OpenAI, Azure OpenAI, Bedrock, Vertex AI, or Ollama)
- **LLM API key** — your embedding provider API key

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For agent memory, the embedding provider is not selectable nor shown to the user.

Comment on lines +516 to +522
RAM automatically deduplicates long-term memories at three levels:

| Level | What it checks | Example |
|---|---|---|
| Exact | Identical text strings | Two extractions producing word-for-word duplicates |
| Semantic | Meaning similarity above a configurable threshold | "User prefers Python" vs "User likes Python" |
| LLM-assisted | Uses an LLM to judge if two memories convey the same fact | "Sarah lives in Austin" vs "Sarah is based in Austin, TX" |

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think this is implemented

Comment on lines +599 to +624

### Forgetting

Long-term memories age and get removed if they are not accessed. This prevents the long-term store from growing unboundedly and keeps retrieved context relevant.

| Config key | Description | Default |
|---|---|---|
| `FORGETTING_ENABLED` | Whether forgetting is active at all | `true` |
| `FORGETTING_EVERY_MINUTES` | How often the forgetting job runs | `1440` (daily) |
| `FORGETTING_MAX_AGE_DAYS` | Maximum age of a memory regardless of access | `30` |
| `FORGETTING_BUDGET_KEEP_TOP_N` | For a given entity/session, keep at most this many memories | `100` |

```yaml
memory:
forgetting:
enabled: true
every_minutes: 1440 # daily
max_age_days: 30
budget_keep_top_n: 100
```

**Tuning guidance:**
- Increase `max_age_days` for users who return infrequently (e.g., monthly support contacts)
- Decrease `budget_keep_top_n` for high-volume apps where you want to limit store size
- Set `enabled: false` if your use case requires permanent retention (compliance scenarios)

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This feature does not exist either

Comment on lines +1007 to +1009
{{< note >}}
There is currently no official Python SDK for the Agent Memory API. All examples use the REST API via `httpx`. The RAM API base URL and credentials differ by deployment:

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There is a python SDK here: https://pypi.org/project/redis-agent-memory/

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants