Kognitos

# Fix list: Kognitos

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/kognitos, the October 2026 research run, assessed 21 October 2026. Grade C, 57 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on Kognitos: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Payments & pricing, 20 out of 100, up to 10 more on the total

Why it scored 20: No x402, MPP or L402 (0 of 40). No published prices. The FAQ says pricing is consumption-based and quoted by sales, and /pricing returns 404 (0 of 20). The FAQ describes a free workspace where anyone can sign up, build and run automations, and the onboarding guide has no card step (20). A person signs up in a browser with an emailed code and creates the token in the app (0). Total 20.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).

- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.
- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login.
- 20, a free tier or trial that doesn't need a card.
- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).

Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.

Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.

## 2. Reliability, 58 out of 100, up to 8.4 more on the total

Why it scored 58: Graded as a hosted service on the REST API and MCP server. status.kognitos.com on incident.io with App and REST APIs components for the US, EU and UK and a history view (20). The history reports no incidents in July, August, September or October 2026 (30). No request-rate limit with numbers was found. The limits page gives only a 15-second request timeout (0). 429 is listed as "Rate limited" with no Retry-After, backoff or safe-retry guidance (3 of 15). The terms FAQ says enterprise SLAs sit in the order form, and none is published (0). The REST API and MCP server aren't marked beta, but the REST and MCP access to Context Graphs announced on 21 October is a customer preview (5 of 10). Total 58.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 3. Agent ergonomics, 53 out of 100, up to 7.6 more on the total

Why it scored 53: The MCP server has 19 tools that group actions, with reads and writes in separate tools, which gives a client a read-only subset (15 + 3 of 25). `page_size` and `page_token` pagination and AIP-160 filters on list endpoints (18 of 20). Errors are listed by HTTP status with one-line meanings, with no error codes an agent could branch on (8 of 20). No idempotency key or retry guidance for invoke or other writes, and MCP tool annotations couldn't be read (2 of 20). One official SDK, `@kognitos/node` 1.0.2 of 10 March 2026, and nearly every path needs `organization_id` and `workspace_id` (7 of 15). Total 53.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 4. Maintenance & community, 45 out of 100, up to 4.8 more on the total

Why it scored 45: Read for a closed service. AI Employees and Context Graphs were announced on 21 October 2026 (30). Fewer than three dated product changes in the 90 days to 21 October were found, the Context Graph launch of 4 August and the 21 October launch, since the docs' Release Notes are empty and the OpenAPI file was last released on 14 April 2026 (0). Support by email on the status page and a sales contact, with no public product changelog (5 of 15). One official SDK, last released 10 March 2026, and no entry in the official MCP registry (5 of 15). The SDK and OpenAPI repositories use semantic-release, and the OpenAPI repository has a Spectral lint config (5 of 10). Total 45.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## 5. Schema & documentation, 71 out of 100, up to 4.7 more on the total

Why it scored 71: Public OpenAPI 3.0.0 file in github.com/kognitos/openapi (MIT), release 1.1.0 of 14 April 2026, 102 operations. It doesn't cover Context Graphs, and the MCP tools' JSON Schemas couldn't be read without a token (22 of 25). llms.txt and a Markdown copy of every docs page (10). The reference describes each operation and field, and the MCP page gives each tool's purpose and actions and warns about untrusted input, with little on when not to use a tool (12 of 20). 41 enums and typed fields, but AIP-160 `filter` strings and free-form `artifacts` maps (10 of 15). cURL quick start with responses and a status-code table, with no error body format (9 of 15). Paths are versioned at `/api/v1` and the OpenAPI repository keeps a semantic-release changelog, but the docs' Release Notes section is empty (8 of 15). Total 71.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## 6. Security & auth, 74 out of 100, up to 4.6 more on the total

Why it scored 74: MCP sign-in is OAuth 2.1 with discovery and short-lived tokens, scopes `mcp:access` and `offline_access`. REST tokens are scoped to chosen workspaces and to All, Read only or Restricted permissions, expire after 7 days to 1 year, and are revoked by deleting them. Rotation is by creating a new key first. No secret in a URL is documented (28 of 30). Read only and Restricted keys and five RBAC roles give least privilege. No confirmation step for destructive MCP actions is documented. AI Employees' per-action approval limits are preview only (15 of 20). The MCP docs warn that the assistant acts on content it reads and advise against connecting it to automations that take untrusted input without review, which is guidance without a mitigation (8 of 15). Org-wide audit log that can't be edited, with the token named as actor for API actions and CSV export (15 of 15). SOC 2 Type II, HIPAA, GDPR and ISO 27001 are claimed on the Context Graph page and as "aligned" in the privacy FAQ. The trust portal is behind a Cloudflare challenge we didn't pass. No security.txt and no bug bounty found (8 of 20). Total 74.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.
- 0 to 15, audit logs or per-call visibility for the operator.
- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.

Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.

## 7. Transparency & trust, 67 out of 100, up to 2.9 more on the total

Made of editorial 47, provenance 86.

Why it scored 67: Closed service with published terms (Published Date 19 November 2025, California law), and MIT for the OpenAPI file, SDK and plugin (15 of 30). The terms limit use of Customer Data to running the service and support, allow Telemetry Data for improvement and allow deletion 15 days after the agreement ends. The terms and privacy FAQs say customer data isn't used to train foundation models and retention follows customer configuration. The privacy notice (11 May 2026) gives retention only as "as long as necessary", and the sub-processor list is on the unreadable trust portal (18 of 30). REST API v1 is marked deprecated with no end-of-life date, and the terms FAQ promises advance email notice of material changes, with no deprecation policy (6 of 20). The status page shows US, EU and UK regions. Sub-processors and AI model providers weren't found in readable pages (8 of 20). Total 47.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).

The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.

Provenance checks not met in full (half of this category, computed from checked facts):

- Domain age: kognitos.com, registered 2020-05-19 (6 years) (11 of 15)
- security.txt: not found (0 of 10)

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- unchecked: the REST and MCP interface for Context Graphs. The release says customers' own agents can connect, but no endpoint, tool or schema is public.
- unchecked: whether the free workspace has AI Employees, Context Graphs, API keys or MCP access, and what its usage limits are. No plan page was found.
- unchecked: the MCP tools' input schemas and readOnlyHint or destructiveHint annotations. The server answers 401 without a token.
- unchecked: the trust portal (SOC 2 report, ISO 27001 certificate, sub-processors). trust.kognitos.com served a Cloudflare challenge.
- unchecked: when the MCP server was first published. The docs page carries no date.
- Vendor claims not checked: "the first neurosymbolic AI platform for business automation", #1 Overall in the 2026 ISG Buyers Guide, the Gartner sample vendor mentions, 23x ROI from a customer quote, and the reliability study cited in the release.
- Own products: Kognitos runs business automations for companies. That isn't LocalGhost's job (local-first personal AI) or letme's (picking the best tool for an agent's task and handling its auth and payments).

## Weaknesses

- AI Employees and Context Graphs are a preview for existing customers, with no public docs for their REST or MCP access
- No price list. Paid use is quoted by sales on consumption
- No numeric rate limit and no retry guidance found. 429 is listed with no Retry-After or backoff guidance
- The docs' Release Notes section is empty, and the OpenAPI file was last released on 14 April 2026
- The trust portal is behind a bot check, so compliance reports and sub-processors couldn't be read

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Don't plan on Context Graph or AI Employee calls. Neither the MCP tool list nor the OpenAPI file has them, and both are customer preview only
- Call `GET /me/organizations` and then list workspaces first, since nearly every path needs `organization_id` and `workspace_id`
- Invoke with `stage: AUTOMATION_STAGE_PUBLISHED` for production and `AUTOMATION_STAGE_DRAFT` to test, then poll the run until it leaves `pending` or `running`
- Treat a `waiting` run as a question for a person. Answer it through the exception tools or ask the owner, rather than retrying
- Use OAuth with callback port 3777 for the MCP server. Another port can fail with a callback URL mismatch

## What the review panel asked for

- confirmation for destructive MCP actions
- an end-of-life date for legacy v1
- a dated product changelog
- numeric rate limits with Retry-After
- idempotency keys for invoke
- Publish a rate card
- State free workspace limits

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.