Skip to content
| Marketplace
Sign in
Visual Studio Code>Machine Learning>Cognee — Project MemoryNew to Visual Studio Code? Get it now.
Cognee — Project Memory

Cognee — Project Memory

cognee

|
1 install
| (0) | Free
Give your editor a persistent, citable memory of the project. Remember and recall knowledge without leaving VS Code, and ask your project memory with answers linked back to source files — powered by Cognee.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Cognee — Project Memory (VS Code)

Give your editor a persistent, citable memory of the project. Remember and recall knowledge without leaving VS Code, and ask your project memory — get answers grounded in what Cognee knows about the current repository, with links back to the source files.

Powered by Cognee. Works against a local Cognee server or a Cognee Cloud tenant.

Features

  • Remember Selection / Remember File / Remember Note — store code, files, or a free-form note into per-repository memory.
  • Ask My Project Memory — a panel that answers questions about the repo, with ranked, clickable citations that open the exact source file.
  • Recall — a quick query from the command palette.
  • Index Workspace — opt-in bulk ingest, respecting .gitignore/.cogneeignore (and always skipping dependency/build directories), with a preflight summary.
  • Forget Project Memory — clear the graph (keep files) or delete the dataset.
  • Per-repository isolation — each repo maps to a stable vscode_<hash> dataset (from the git remote, or the workspace path as a fallback).

Requirements

A reachable Cognee backend — either of:

  • Cognee Cloud (no local setup): your tenant URL (e.g. https://<tenant>.cognee.ai) and an API key.
  • Local server: a Cognee server running on http://localhost:8011 (see the Cognee docs); this path needs its own .env (e.g. LLM_API_KEY).

The extension itself needs no environment variables — it is configured entirely through VS Code settings and secret storage.

Quick start

  1. Run Cognee: Set Up from the command palette.
    • Enter your endpoint (http://localhost:8011 or your Cloud tenant URL).
    • Enter your API key if using Cloud (stored securely in the OS keychain, not in settings).
    • The command runs a health check so you know the connection works.
  2. Open a file, select some code, and run Cognee: Remember Selection.
  3. Run Cognee: Ask My Project Memory and ask a question — the answer appears with its sources.

Commands

Command What it does
Cognee: Ask My Project Memory Open the panel and query the repo's memory with citations.
Cognee: Recall One-off query from the palette.
Cognee: Remember Selection Store the current selection (or the whole file).
Cognee: Remember File Store a file (also on the explorer context menu).
Cognee: Remember Note Store a free-form note typed into an input box.
Cognee: Index Workspace Bulk-ingest eligible files after a preflight confirmation.
Cognee: Forget Project Memory Clear memory (keep files) or delete the dataset.
Cognee: Set Up Configure endpoint + key and run a health check.

Settings

Setting Default Description
cognee.endpoint http://localhost:8011 Cognee backend base URL.
cognee.apiKey "" API key (X-Api-Key). Prefer Cognee: Set Up, which uses secret storage.
cognee.datasetOverride "" Fixed dataset name; empty derives vscode_<hash> per repo.
cognee.searchType auto Recall strategy; auto lets Cognee route the query.
cognee.topK 15 Max recall results.
cognee.includeReferences true Attach source citations to answers.
cognee.ingestion.respectGitignore true Skip .gitignore/.cogneeignore files when indexing (dependency/build dirs are always skipped).
cognee.ingestion.maxFileSizeKb 512 Skip files larger than this when indexing.
cognee.requestTimeoutMs 300000 HTTP request timeout.

How it works

The extension talks to Cognee over the HTTP API:

  • Ask / Recall → POST /api/v1/recall with include_references: true, scoped to the workspace dataset. Citations are parsed from the answer's Evidence block (chunk/document level today), then ranked by relevance to the query and de-duplicated so each source file appears once. Everything the extension ingests carries a Source: <path> header, so a citation resolves straight to the exact file — even when several files share a name. It falls back to a per-workspace path index (the exact path of each file you remember), then snippet-content matching, then a user pick, only when genuinely ambiguous.
  • Remember / Index → POST /api/v1/remember (ingest + graph build in one call).
  • Forget → POST /api/v1/forget.

All editor-agnostic logic lives in src/core (no vscode imports), so it is unit-tested against a mocked backend and can back other editors (e.g. a JetBrains sidecar) unchanged.

Development

npm install
npm run build      # bundle to dist/extension.js (esbuild)
npm run typecheck  # tsc --noEmit
npm test           # vitest — runs against a mocked backend, no live keys

Press F5 in this folder to launch an Extension Development Host with the extension loaded.

License

Apache-2.0. See LICENSE.

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft