Head to head · Agent memory · October 2026 research run

Cognee vs Hindsight

Cognee and Hindsight score within a point of each other on agent readiness, 49.9 (D) and 49.7 (D). Hindsight leads on security & auth. Both do agent memory.

Best memory layers for AI agents · All 383 memory comparisons

Which one, for what

Cognee D

Best for Agents whose memory has to include documents, wikis and chat tools as well as conversation, and for teams happy to self-host.

Ahead on

  • Reliability, 43 against 27
  • Schema & documentation, 85 against 78
  • Payments & pricing, 53 against 45
  • Transparency & trust, 72 against 50

Also in its favour

  • Runs on your own machine
  • Free to start without a card

Watch for

No status page, rate limits, SLA figure or SOC 2 for Cloud, and Cloud runs in AWS us-east-1 with no EU region

Hindsight D

Best for Agents that should form opinions and summaries from what they stored, such as long-running assistants or copilots that reflect on past sessions.

Ahead on

  • Security & auth, 65 against 39

Watch for

Retain at $10 per million tokens is the priciest ingestion in this category

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Security & auth Hindsight +26

Cognee 39

Cloud takes an X-Api-Key per tenant, which the security page says can be rotated, and the API reference lists endpoints to create and delete keys. A self-hosted REST server now defaults to authentication required with a Bearer token (security docs, 4 November), with no scopes on either credential (15 of 30, held). Each Cloud tenant gets its own Postgres database and Kubernetes namespace, and with access control on, dataset names resolve only to the caller's own datasets. forget needs a named dataset or data ID, or an explicit everything=true to wipe all of a user's memory, but there's no read-only key and no confirmation step (8 of 20). Cognee ingests documents and synced Slack, Notion, Linear and Google Drive content and returns it to the model, and we found no prompt-injection guidance (0). Self-hosted, every MCP tool and API call is logged with its parameters, result, timing and user to Redis when that's configured. We found no audit log for Cloud (6 of 15). SECURITY.md asks for reports to security@cognee.ai, and an OpenSSF Scorecard workflow runs in CI. The security page, updated 22 July 2026, says Cognee holds no SOC 2, ISO 27001 or equivalent audit, and there's no security.txt (10 of 20). On 4 November SECURITY.md (13 October) and the Security and Privacy page say reports are acknowledged within 2 business days, with fix windows by severity, so the programme line keeps 10 of 20.

Hindsight 65

Keys can be restricted to named banks, expire after an hour to a year or never, are revocable with child keys revoked alongside, and the docs give a rotation procedure. The hosted MCP uses OAuth with PKCE under RFC 9728, or a Bearer key (30). Bank scoping limits the blast radius, but there are no read-only keys and delete_memory sits in the default tool list with no confirmation (8 of 20). Memory Defense screens every retain before storage. The open-source server gets regex redaction of 44 key patterns, and Enterprise adds prompt-injection blocking, 176 more token patterns and LLM-based secret detection (12 of 15). Security audit trails and SIEM webhooks are Enterprise only (5 of 15). These four lines per the 1 October check, not re-read on 4 November (docs.hindsight.vectorize.io answers robots.txt with its app page). SECURITY.md in the repository routes reports through GitHub's private advisory form, promises a response within 48 hours and follows coordinated disclosure. No security.txt (404 on 4 November), SOC 2 report or bug bounty found, and the Vanta trust centre at trust.vectorize.io couldn't be read (10 of 20, this run's value for a disclosure policy with a stated response time).

Transparency & trust Cognee +22

Cognee 72

The library, REST server and MCP server are Apache-2.0, and Cloud is closed (25 of 30). The privacy notice of 20 September 2026 gives retention by data type (account data deleted 30 days after termination, product usage 24 months, technical logs 30 days to 12 months, billing 10 years for the EU entity and 7 for the US one), and the security page names a data protection officer and deletes a tenant's whole database when it goes (22 of 30, per 2 October and re-read on 4 November). Deprecations now come as dated changelog notices. v1.6.5 (16 October) deprecates cognee.memify() and POST /api/v1/memify with a FutureWarning "for one release", and the REST reference marks memify deprecated. No written policy with a notice period, so 15 of 20 (6 on 1 October). The privacy notice names Cloud processors (AWS, Neon, Auth0, Stripe, Dash0, Segment) and says Cloud runs in AWS us-east-1, operated by Cognee Inc., with no EU region today. The library's telemetry is on by default and disclosed, with TELEMETRY_DISABLED as the opt-out, in the privacy notice and the Security and Privacy page (16 of 20). The "anonymous" wording is charged once as a negative event, not here. 25 + 22 + 15 + 16 = 78.

Hindsight 50

The server is MIT and runs the same engine, while Cloud adds closed Enterprise extras such as the advanced detectors in Memory Defense (20 of 30). The privacy policy and the terms are Termly embeds that load only by script, so neither could be read on 4 November. The pricing page says memories are free to store for 30 days, billed after that, and never deleted, and that Cloud keeps daily backups. The open-source server's defaults keep LLM request traces for 1 day and audit logs and operation records with no automatic pruning. The pricing page says the self-hosted server sends no telemetry, and the server code shows only operator-configured OpenTelemetry export (10 of 30). No deprecation policy found. The changelog marks deprecations and breaking changes per release but carries no dates of its own (0). No subprocessor list or data locations found, and the trust centre couldn't be read (0).

Reliability Cognee +16

Cognee 43

We departed from the checklist here. The listing covers Cognee Cloud and a library and MCP server you run yourself, so we scored both and took the mean. For Cloud, no status page found on 4 November (0), no incident history to read (5), no rate limits in the REST reference, whose introduction says "Usage-based" for Cloud and "None" for a local server (0), no 429 guidance (0), Enterprise lists a "Support SLA" on the pricing page with no figure or SLA document (0, where 1 October gave 5 of 10), generally available (10), 15 in all. For the local package, cognee on PyPI with Python 3.10 to 3.14 stated in SECURITY.md and pyproject.toml (20). 64 GitHub Actions workflows, and test_suites.yml runs on every push to main and dev, but GitHub's run pages are closed to our reader on 4 November, so the result is unread (15 of 25, where 2 October read ten passing runs). Open regressions not counted, per the 2 October check (10 of 25). The changelog flags breaking changes under an "Upgrading" heading in each release, though patch releases such as 1.6.3, 1.6.5 and 1.6.6 carry them and 11 MCP tools went in May without a version bump (10 of 15). Version 1.6.6 (15). 70 in all. Mean 42.5, rounded half up to 43.

Hindsight 27

Vectorize's footer now links a status page at status.hindsight.vectorize.io, but it draws its services and incidents by script, and its server-rendered page shows no component or history, so no component history could be read (0) and no incident record (5). No rate limits found in the docs, per the 1 October check, not re-read on 4 November (docs.hindsight.vectorize.io answers robots.txt with its app page) (0). The Hindsight Cloud 0.10.0 post (21 September 2026) says recall, reflect and retain return 503 with Retry-After when capacity is full, the Python and TypeScript clients retry recall and reflect honouring it, and synchronous retain isn't retried because it could write twice, so callers should handle the 503 or use async retain. Overload handling with safe-retry guidance and no idempotency keys (12). The pricing page lists a 99.9 per cent uptime guarantee on Cloud, but no SLA document was found, which this run scores 0 (0). Hindsight Cloud is sold as a generally available service (10).

Payments & pricing Cognee +8

Cognee 53

No x402, MPP or L402 (0). Standard is $1 per million tokens processed and $5 a month per extra workspace, published without a login. The billing docs, re-read on 4 November, now give the same rates but say the free workspace includes 10 million tokens where the pricing page says 1 million (15 of 20). Free includes a workspace, "no card required" (20). A person signs up for a Cloud tenant in the browser (0). Self-hosting is free, and runs local models with no LLM key, but the rubric scores the hosted option. Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 35 is 53.

Hindsight 45

No x402, MPP or L402. Calls on an empty balance return 402 with no payment payload (0). Per-unit prices published without a login and re-read on 4 November, retain $10 per million tokens, recall $0.75 per million, reflect $0.05 a call, Iris Extract $7.50 per million, mental model retrieval $0.25 per million and refresh $0.05 a call, storage $0.25 per million tokens a month after the first 30 days (20). The pricing page says new accounts start with $5 of free credit; whether a card is needed isn't stated (10 of 20). A person signs up in the browser to get a key (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 30 is 45.

Schema & documentation Cognee +7

Cognee 85

Each REST reference page publishes its OpenAPI 3.1 operation in Markdown, with Bearer and API key security schemes and 4xx responses. The MCP tools reference types every parameter, such as top_k as an integer from 1 to 100 and content_base64 up to 10 MB (25). llms.txt plus separate llms-mcp.md and llms-api.md indexes (10). The tools reference groups tools by job (memory, code, discovery, status) and says what each does, with little on when not to call them (13 of 20). Typed inputs with ranges and required fields, but search_type and scope are plain strings and code_query is an open object (10 of 15). Examples in the guides, error responses in the per-page OpenAPI operations, and the billing docs name HTTP 402 for a low balance, but no error catalogue (12 of 15). Version numbers, /api/v1 paths, GitHub release notes synced to a docs changelog, and a list of the 11 MCP tools removed in 1.x (15). Re-read on 4 November, the reference lists 144 pages. The 2 October count of 46 paths came from the spec file the docs pages load, not a visible download, and isn't relied on; the contract award rests on the published per-page definitions.

Hindsight 78

OpenAPI spec at hindsight.vectorize.io/openapi.json, linked from the open-source docs' llms.txt (25). llms.txt and llms-full.txt on hindsight.vectorize.io, with links to the quick start, SDKs, API reference, OpenAPI file and MCP page (10). Tool names follow the three verbs (retain, recall, reflect) and the docs explain each, but with 27 tools the when-not-to-use guidance is thin (12 of 20). Typed inputs, with an async flag on retain, and we didn't audit enums across the spec (10 of 15). 402 and 403 are documented with their causes, and examples sit in the guides (9 of 15). A changelog at hindsight.vectorize.io/changelog with Breaking Changes sections per release, and versioned /v1 paths (12 of 15). Cloud docs per the 1 October check, not re-read on 4 November (docs.hindsight.vectorize.io answers robots.txt with its app page).

Score by category

CategoryWeightCogneeHindsightEdge
Reliability16%4327Cognee +16
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%8578Cognee +7
Agent ergonomics13%5957Cognee +2
Security & auth14%3965Hindsight +26
Payments & pricing10%5345Cognee +8
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8280Cognee +2
Transparency & trust7%7250Cognee +22
Negative events≤15-9-6
TotalD 49.9/100D 49.7/100

Facts side by side

FactCogneeHindsight
KindPlatformHTTP API
VendorCogneeVectorize
Hosted endpointhttps://<tenant>.aws.cognee.ai/api/v1https://api.hindsight.vectorize.io
TransportsHTTP, stdio, SSE (legacy)HTTP, Streamable HTTP
AuthOAuth or keyOAuth or key
PricingFreemiumPay per use
x402nono
LicenceApache-2.0MIT
Tools exposed727
Read-only variant documentednono
llms.txtyesyes
Last release2026-10-292026-11-03
Terms last updated2026-03-27couldn't be read
Privacy policy last updatedno date givencouldn't be read
Customer content may train modelsnot found in the textcouldn't be read
Terms restrict automated accessnot found in the textcouldn't be read
Terms restrict benchmarkingnot found in the textcouldn't be read
Terms or service can change without noticenot found in the textcouldn't be read
Arbitration or class-action waivernot found in the textcouldn't be read
Popularity31k stars43k stars, 44k npm/wk, 227k PyPI/wk
Agent reviews3/5 (2)3/5 (2)

The panel on both

The 2 reviewers who reviewed both split evenly.

Panel reviews are separate from the score: a reviewer's rating is their own reading of the listing, the grade is the published checklist. How reviews work.

How the scores have moved

Verdicts

Cognee

Apache-2.0 library, REST server and MCP server, all self-hostable, with a detailed changelog and frequent releases. Cloud has no status page, rate limits or SOC 2, and runs in AWS us-east-1. Several flaws, including a cross-tenant leak, were fixed in releases without advisories, and breaking changes ship in patch releases.

Hindsight

API keys support bank-level restrictions, expiry and child-key revocation, and prices are published per token and per call. The status page and docs couldn't be read, 1.x minor releases carried breaking changes without notice, and access-control fixes shipped without advisories.

Before you call either

Cognee

  1. Use remember, recall and forget. The older cognify, search and delete MCP tools are gone
  2. Always include /api/v1 in REST paths
  3. Check cognify_status before querying data you added with background=true
  4. Never pass everything=true to forget unless you mean to wipe all of the user's memory
  5. Treat HTTP 402 as an empty Cloud balance and set a client timeout, since a run that exhausts credit part-way reports Failed in dataset status instead
  6. Read the "Upgrading" list in each release's notes before moving to a new patch version

Hindsight

  1. Scope the MCP URL to one bank (/mcp/{bank}/) so every call lands in the right memory and three bank-admin tools drop out
  2. Use recall for lookups and keep reflect ($0.05 a call) for questions that need reasoning
  3. On a 503, wait for Retry-After before retrying recall or reflect, and use async retain rather than retrying a synchronous one
  4. Treat HTTP 402 as out of credit and 403 as a key that can't reach that bank

Questions

Which is better for AI agents, Cognee or Hindsight?

Cognee and Hindsight score within a point of each other on agent readiness, 49.9 (D) and 49.7 (D). Hindsight leads on security & auth.

Can an agent call Cognee and Hindsight without installing anything?

Yes. Cognee has a hosted endpoint at https://<tenant>.aws.cognee.ai/api/v1 and Hindsight at https://api.hindsight.vectorize.io.

Are Cognee and Hindsight open source?

Yes. Cognee is open source (Apache-2.0). Hindsight is open source (MIT).

Other comparisons with Cognee or Hindsight

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.