apps/docs/integrations/memory-graph.mdx and apps/memory-graph-playground/README.md
documented an older version of the @supermemory/memory-graph public API. Following
the Quick Start, Props Reference, or Exports section as written produced code that
either fails to type-check or silently drops props the component no longer has.
Verified against packages/memory-graph/src/{index.tsx,types.ts,api-types.ts,mock-data.ts,
constants.ts} and cross-checked the API normalization logic against two independent
real call sites: apps/web/components/memory-graph/hooks/use-graph-api.ts and
apps/memory-graph-playground/src/app/page.tsx.
- Exports section listed Legend, NodeDetailPanel, SpacesDropdown, useGraphInteractions,
colors, LAYOUT_CONSTANTS - none of these are actually exported. Corrected to the
real exports and added the previously-undocumented ./mock-data subpath.
- All examples used loadMoreDocuments/totalLoaded/selectedSpace/onSpaceChange/
showSpacesSelector, none of which exist on MemoryGraphProps. Corrected to
onLoadMore/totalCount and removed the Controlled Space Selection example
(that feature no longer exists on the component).
- documents was typed as DocumentWithMemories[] with a fabricated shape; the real
prop type is GraphApiDocument[]. Added a toGraphDocument mapping example grounded
in the real production code that talks to /v3/documents/documents.
- Rewrote Props Reference and Data Types to match the real MemoryGraphProps and
GraphApiDocument/GraphApiMemory interfaces field-for-field, including several
real props that weren't documented at all before.
- Removed the false 'space selector visible/hidden' claim from the Variants section
(no such component exists anywhere in packages/memory-graph/src); kept the 0.8x/0.5x
zoom figures, which checked out against constants.ts.
- Made the Pages Router and Express tabs in Backend API Route self-contained (each
defines its own toGraphDocument mapping) instead of depending on code shown only
in the App Router tab, since CodeGroup tabs are alternatives a reader copies
individually.
- Applied the identical corrections to apps/memory-graph-playground/README.md,
which had the same drift independently.
Address Graphite automated review feedback:
- Set the content field in all three toGraphDocument mappings (was a valid but
unpopulated field on GraphApiMemory). Confirmed via use-graph-data.ts that it's
inert for rendering (always overwritten with mem.memory before draw), so this
wasn't visible data loss, but it's a free fix that matches the type exactly.
- Replaced the (data.documents as RawDocument[]) cast with a proper
RawDocumentsResponse type annotation on data in the App Router and Pages Router
tabs, per the custom TypeScript style rule Graphite flagged. Implemented what
the comment's prose described rather than its literal suggested diff, which
would have just deleted the cast and left data as any.
Docs-only change, no .ts/.tsx files touched. This repo's biome.json has no
Markdown/MDX support configured, so format-lint doesn't apply here; checked
fence and MDX-component balance by hand instead (34 fence lines, all CodeGroup/
Note/Warning/Card tags balanced 1:1).
Updates the Supermemory MCP docs for the revamped tool, space, widget, and OAuth flows.
- adds a screenshot-backed ChatGPT Web setup guide with light and dark variants
- refreshes the overview, setup, tools, spaces, and widget docs
- keeps manual JSON configuration in a dropdown
Tested all four local MCP docs routes and image references.
for mcp image light mode -> dark mode images and dark mode has light mode (since the background was blending in i made this way to refocus user attention on the screenshots)
Keep docs asset URLs repo-root-relative and pass images as literal MDX children so Mintlify can compile them through OptimizedImage under the /docs mount.
Mintlify's official guidance says: “Image paths are root-relative from your docs repository.” It also says relative paths such as `./screenshot.png` are unsupported. See [Image embeds](https://www.mintlify.com/docs/create/image-embeds).
The existing `/images/...` paths were correct. The failure came from passing them through custom-component string props, which kept Mintlify from seeing those images during its MDX transform.
- fix hero, building-block, and Slack avatar images
- preserve Slack avatar clipping
- normalize Hermes and Company Brain icon sizing
Tested with Mintlify validation, broken-link checks, and browser checks across every changed route.
Fixes broken docs navigation and asset paths from the site audit.
- serves docs images from `/images` and adds the missing Cartesia icon
- corrects homepage, console, and LinkedIn destinations
- removes agent-only comparison headings from the web TOC
Tested with Mintlify validate and Mintlify broken-links.
## Docs: add Forget Matching endpoint + fix stale Forget Memory docs
### What this does
- **Adds docs for the new** **`POST /v4/memories/forget-matching`** **endpoint** — semantic/promptable mass-forget. Covers `dryRun` (preview), `threshold`/`maxForget` safety bounds, the request/response shape, and `forgetBatchId`.
- **Corrects the existing "Forget Memory" section** to match the actual implementation.
### ⚠️ No API surface changed
This PR is **docs-only**. The existing forget endpoint's behavior/contract is untouched — the previous docs were simply **wrong** and described a route that has never existed:
| | Old docs (incorrect) | Actual implementation (unchanged) |
| --- | --- | --- |
| Method + path | `POST /v4/memories/{id}/forget` | `DELETE /v4/memories` |
| Body | — | `{ id \| content, containerTag, reason? }` |
The handler (`forgetMemory` in `apps/api/src/routes/v4/memories/handlers.ts`) was not modified — this just makes the docs reflect reality.
### Also
- Small accuracy cleanups (response field descriptions, realistic example IDs).
### TL;DR
Adds documentation for the Memory Review endpoints and Profile Buckets feature.
### What changed?
**Memory Review (`memory-review.mdx`)**
- Added a new documentation page covering the two inferred memory review endpoints: `GET /v3/container-tags/{containerTag}/inferred` and `POST /v3/container-tags/{containerTag}/inferred/{memoryId}/review`.
- Documents the three review actions (`approve`, `decline`, `undo`) and how each affects search ranking and memory state (`isInference`, `isForgotten`, `reviewStatus`).
- Includes request/response examples in both `fetch` and cURL, a field reference table, error codes, and a collapsible React Query hooks example for building a review UI.
- Registered the new page in `docs.json` under the "Manage Content" group and linked to it from the Memory Operations next steps.
**Profile Buckets (`user-profiles.mdx`)**
- Added a "Profile Buckets" section explaining custom topical categories (`preferences`, `goals`, `work`, etc.) as a complement to `static`/`dynamic` profile sections.
- Documents the `include`, `buckets`, and `filters` query parameters on the profile endpoint.
- Covers the `GET /v4/profile/buckets` endpoint for listing configured bucket definitions, with request/response examples and a field reference.
- Explains the `[Recent]` / `[Summary]` label convention used in bucket and dynamic profile entries.
- Updated the `ProfileResponse` TypeScript interface to mark `static` and `dynamic` as optional and add the `buckets` field.
### How to test?
- Navigate to the docs site and confirm "Memory Review" appears in the sidebar under "Manage Content".
- Verify all code examples render correctly and tabs switch between `fetch` and cURL variants.
- Confirm the React Query accordion expands and displays the TypeScript snippet.
- Check that the Profile Buckets section renders inline within the User Profiles page, including the response JSON blocks and the tip/note callouts.
### Why make this change?
Inferred (derived) graph memories are down-weighted in search until reviewed, but there was no documentation explaining how to surface or act on them. Similarly, profile buckets were a shipped feature with no public-facing docs. These additions give developers the reference material needed to build review UIs and use topical bucket filtering in their integrations.
## Summary
- move Granola from Max-gated to Pro-gated in Nova integrations and add-connection flows
- add Granola to Pro plan card connector copy in billing
- update Granola connector docs to say Pro Plan or higher
## Testing
- bunx biome check apps/web/components/settings/billing.tsx apps/web/components/integrations-view.tsx apps/web/components/add-document/connections.tsx apps/docs/connectors/granola.mdx
- git diff --check
Note: onboarding-brain was intentionally left unchanged.
## Summary
- Rename the Claude Code plugin docs references from `claude-supermemory` / `Claude-Supermemory` to `supermemory`
- Update install and command examples to use `/plugin install supermemory` and `/supermemory:logout`
## Testing
- Ran `git diff --check`
---
**Session Details**
- Session: [View Session](https://supermemory.us1.vorflux.com/agent-sessions/cc17c7ae-0fb2-4e2c-bb4c-20a66900d720)
- Requested by: Sreeram Sreedhar (sreeram@supermemory.com)
- Address comments on this PR. Add `(aside)` to your comment to have me ignore it.
### TL;DR
Documents the new `filterByMetadata` parameter for the memory ingestion API, added in [supermemoryai/mono#1283](https://github.com/supermemoryai/mono/pull/1283).
### What changed?
- Added a new "Filtered Writes" section to the `add-memories.mdx` page explaining how to scope memory context during ingestion
- Added `filterByMetadata` to the Parameters table with a link to the new section
- Included TypeScript, Python, and cURL examples
- Documented scalar vs array value matching semantics (AND/OR logic)
### Key documentation points
- The metadata itself is still written to the document, but memories are only built on top of existing memories matching the filter
- Scalar values match exactly, array values create OR conditions
- Multiple keys are combined with AND logic
### Related
- Implementation PR: [supermemoryai/mono#1283](https://github.com/supermemoryai/mono/pull/1283)
---
**Session Details**
- Session: [View Session](https://supermemory.us1.vorflux.com/agent-sessions/14d33783-50b0-4fc8-8e0f-abc6346336f1)
- Requested by: Dhravya Shah (dhravya@supermemory.com)
- Address comments on this PR. Add `(aside)` to your comment to have me ignore it.
- Published the v2.0.0 docs and a 1.4 → 2.0 migration guide so existing users have a clear upgrade path.
- Updated the four integration pages (AI SDK, OpenAI, Mastra, VoltAgent) to reflect v2 defaults and link to the migration guide.
- Added a short explainer on the two required fields (containerTag, customId) so new users aren't blocked at first integration.
**`withSupermemory`** **(AI SDK)**
- **`skipMemoryOnError`** **defaults to** **`true`**. memory errors/timeouts log and the model runs on the **original** prompt unless you set `skipMemoryOnError: false`.
- **Pre-LLM** **`/v4/profile`** **is aborted after 5s** via `AbortSigna`
**Docs**
- `packages/tools/README.md`, **`apps/docs/integrations/ai-sdk.md`**
### TL;DR
Added Python SDK for integrating Supermemory with Cartesia Line voice agents, enabling persistent memory capabilities.
### What changed?
Created a new Python SDK package (`supermemory_cartesia`) that provides:
- `SupermemoryCartesiaAgent` wrapper class that enhances Cartesia Line agents with memory capabilities
- Memory retrieval and storage functionality that integrates with the Supermemory API
- Utility functions for memory formatting, deduplication, and time formatting
- Custom exception classes for error handling
- Comprehensive documentation and type hints
The implementation includes:
- Memory enrichment for user queries
- Automatic storage of conversation history
- Configurable memory retrieval modes (profile, query, full)
- Background processing to avoid blocking the main conversation flow
### How to test?
```python
from supermemory_cartesia import SupermemoryCartesiaAgent
from line.llm_agent import LlmAgent, LlmConfig
import os
# Create base LLM agent
base_agent = LlmAgent(
model="gemini/gemini-2.5-flash-preview-09-2025",
config=LlmConfig(
system_prompt="You are a helpful assistant.",
introduction="Hello!"
)
)
# Wrap with Supermemory
memory_agent = SupermemoryCartesiaAgent(
agent=base_agent,
api_key=os.getenv("SUPERMEMORY_API_KEY"),
user_id="user-123",
)
# Use memory_agent in your Cartesia Line application
```
### Why make this change?
This SDK enables Cartesia Line voice agents to maintain persistent memory across conversations, enhancing user experience by:
1. Providing contextual awareness of past interactions
2. Remembering user preferences and important information
3. Reducing repetition in conversations
4. Creating more personalized and natural voice interactions
The integration is designed to be lightweight and non-blocking, ensuring that memory operations don't impact the responsiveness of voice interactions.
Single-page changelog covering Feb 2024 through Mar 2026 with filterable
tags (API, SDK, Console, MCP, CLI, Integrations). Replaces the split
overview/developer-platform pages. Adds redirect for old URL.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
