Manus
Spectrum 0.2 beta · Public tier · read 2026-11-10 · not reviewed by its vendor yet
Manus AI's general agent that plans and carries out tasks for a user in a cloud sandbox, and also through a desktop app, a browser extension and an API.
Standout. Autonomy 3 (Acts): "tools that the current Manus agent runs in the main task loop"
Next unlock. Control 2 (Configurable): public evidence that what it can reach is enumerated: scopes, permissions, tools would get there.
Trust. FACT-Ray: 18 facts: 18 π Moderate · found by an agent, captured and hashed, checked by 2 other agents (same model family, shared repository); no human has signed it.
Standards constellation. 52 provisions lit across 8 of 8 instruments.
Evidence: 12 sources, captured and hashed · the card's page has the vault's read key · Not reviewed by its vendor yet · correct or claim this card
The walk down from this card: each dimension, the indicators its level stands on (a link to the indicator's row), the public source of each answer (a link out; dashed means available, not default) the instruments its provisions sit in (a link to the instrument) and each provision itself (a link to its own page). A solid frame is evidenced; a dashed frame is not yet evidenced, which is not a no.
What each level rests on
Autonomy: what does A3 Acts rest on? 2 facts
I08 It executes actions itself
I17 The agent handles routine tasks; the person is consulted on exceptions
Reach: what does R2 Integrated rest on? 2 facts
I10 It works across several domains or systems
I11 It connects to external tools or services across a trust boundary
Control: what does C1 Stated rest on? 1 fact
I38 High-impact actions need a person's approval
Openness: what does O2 Documented rest on? 2 facts
I22 Its limits and known failure cases are documented for those who use it
I58 The data it uses is described
Each path has its verbs on the edges; every node is a link, the source opens the page, a dashed box is an open question, not a no.
This card rates a product as its vendor documents it, not anyone's deployment of it. A feature the operator has to turn on counts, and is marked as available. Levels are computed from the evidence below by published rules. A dimension with no evidence reads "not yet evidenced".
Autonomy: level 3, Acts
What does the agent do on its own? Completes routine tasks end to end; a person handles exceptions. This level holds with: Manus Studio, ask for approval; Manus Studio, working locally. As shipped, with nothing turned on, it is not yet evidenced.
I08It executes actions itself. The API docs list actions the agent carries out in its task loop, such as sending email, changing calendar events, running terminal commands and deploying a service. This is the API's documentation of the task loop. "tools that the current Manus agent runs in the main task loop" Source (read, documented, captured 2026-11-10T04:59Z, sha256b51a0a23e9bc2981) · π Moderate · S1·R5·A1 ×3I17The agent handles routine tasks; the person is consulted on exceptions. In Manus Studio the default Approve for me mode auto-approves safe local actions and asks the person when a decision is needed. "Reviews and auto-approves safe local actions." Source (read, documented, available, captured 2026-11-10T04:59Z, sha25696079a92f854e6e0) · π Moderate · S1·R5·A1 ×3
This evidence touches 4 provisions: Salesforce, Microsoft, EAAMM.
Next unlock: level 4, Orchestrates. Public evidence that the agent works independently; the person approves critical decisions only.
Reach: level 2, Integrated
How connected is it? Documented connectors to common tools.
I10It works across several domains or systems. Connectors let the agent read data from and act in apps such as Gmail and Notion once the user connects them. "lets Manus read data from an app, such as Gmail or Notion" Source (read, documented, available, captured 2026-11-10T04:59Z, sha256f1932af9dccead99) · π Moderate · S1·R5·A1 ×3I11It connects to external tools or services across a trust boundary. The Similarweb data integration is used without setup, the agent fetching metrics from the external service. "Manus automatically connects to Similarweb and retrieves the exact metrics" Source (read, documented, captured 2026-11-10T04:59Z, sha256f44644af396f1ad0) · π Moderate · S1·R5·A1 ×3
This evidence touches 8 provisions: Salesforce, Microsoft, OWASP, NIST AI RMF.
Next unlock: level 4, Ecosystem. Public evidence that agents operate across organisational boundaries or disparate stacks; and that others build on it: third parties publish integrations, skills or agents for it.
Control: level 1, Stated
Can people stay in command? Documents say how people approve, override or stop it. This level holds with: Through the API. As shipped, with nothing turned on, it is not yet evidenced.
I38High-impact actions need a person's approval. The API's confirmation events list action types that wait for the user before they run, among them sending email, deleting calendar events, deploying a service and running a command; the page is for integrators who build their own confirmation interface, does not say when the waits occur in the app, and some types accept global_allow, which skips later ones. "Send, save as draft, or cancel" Source (read, documented, available, captured 2026-11-10T04:59Z, sha256b51a0a23e9bc2981) · π Moderate · S1·R5·A1 ×3
This evidence touches 6 provisions: OWASP, EU AI Act, AAGMM, NIST AI RMF, Salesforce, Microsoft.
Next unlock: level 2, Configurable. Public evidence that what it can reach is enumerated: scopes, permissions, tools.
Openness: level 2, Documented
Does it show its work? Public documents on how it works, its data, its models and its limits. This level holds with: Browser Operator. As shipped, with nothing turned on, it is level 1, Disclosed.
I22Its limits and known failure cases are documented for those who use it. The Browser Operator page lists limitations, including complex interactions and sites with anti-bot measures. "Complex interactions, such as drag-and-drop and multi-step forms, may not work perfectly." Source (read, documented, available, captured 2026-11-10T04:59Z, sha2569c63bdb4aab197d1) · π Moderate · S1·R5·A1 ×3I58The data it uses is described. The sandbox post lists what the task sandbox stores, and the Studio page says what is sent to the cloud. "Your Manus Sandbox stores the files needed during task execution, including" Source (read, documented, captured 2026-11-10T04:59Z, sha25633be308ad9b32242) · π Moderate · S1·R5·A1 ×3
This evidence touches 7 provisions: NIST AI RMF, EU AI Act, AAGMM, EAAMM, Salesforce.
Next unlock: level 3, Measured. Public evidence that test or evaluation results are published.
How far to trust this card
Status: Draft. Draft means the facts were found in public pages and reviewed by agents. The company has not confirmed them and no customer has gone on record. FACT-Ray says how far each fact can be trusted: who says it (Source), how firmly it is recorded (Record) and who stands behind it (Accountability). It never changes a level.
- The facts: 18 facts: 18 π Moderate. Each is written beside its evidence above.
- The chain: found by an agent, captured and hashed, checked by 2 other agents (same model family, shared repository); no human has signed it.
- Who: produced by
gamma.studio@web-summit.dinis; checked bygamma.review@web-summit.dinisandgamma.approve@web-summit.dinis. The actors. - What would raise it: the company's own file at
/.well-known/fact-ray.json(Source 2), a customer's statement of use (Source 3), or a named person signing the card (how).
The standards constellation
One piece of public evidence can speak to several frameworks at once. "Touched" never means "complies". The share is of the provisions that can be read from outside.
| Instrument | Provisions touched | Of those publicly readable |
|---|---|---|
| Salesforce | 11 | 24 |
| EAAMM | 3 | 8 |
| OWASP | 7 | 18 |
| MCP authorization | 2 | 8 |
| AAGMM | 4 | 6 |
| Microsoft | 5 | 23 |
| NIST AI RMF | 15 | 20 |
| EU AI Act | 5 | 6 |
The 52 provisions lit
Each line item, with the facts that reach it. "Evidenced" is a link of the kind same: the fact is what the provision asks. "Partly" is a link of the kind part: the fact is one piece of it. "Its absence" points the other way. Each provision has its own page with its text and its place in the source.
| Provision | Instrument | Reached by | How far |
|---|---|---|---|
sf.a04 Level 0, how companies can advance from Level 0 to Level 1 | Salesforce | I58 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
sf.a06 Level 0, how companies can advance from Level 0 to Level 1 | Salesforce | S1 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
sf.a07 Level 1, how to advance from Level 1 to Level 2 | Salesforce | I08 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
sf.a08 Level 1, how to advance from Level 1 to Level 2 | Salesforce | I58 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
sf.a10 Level 1, how to advance from Level 1 to Level 2 | Salesforce | I57 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
sf.L2 Level 2: Simple Orchestration, Single Domain | Salesforce | I08 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
sf.a16 Level 2, how to advance from Level 2 to Level 3 | Salesforce | I10 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
sf.L3 Level 3: Complex Orchestration, Multiple Domain | Salesforce | I10 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
sf.a17 Level 3, how to advance from Level 3 to Level 4 | Salesforce | I10 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
sf.a22 Level 3, how to advance from Level 3 to Level 4 | Salesforce | I38 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
sf.a28 Level 4, how to maximize value | Salesforce | I57 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
eaamm.L2 L2 Collaborative (survey item Q9) | EAAMM | I16 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
eaamm.L3 L3 Consultative (survey item Q9) | EAAMM | I17 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
eaamm.D2 Readiness dimension 2 | EAAMM | I58 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
owasp.AT4 AT4 Code Executing Agent | OWASP | I04 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
owasp.AT6 AT6 Externally Extended Agent | OWASP | I11 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
owasp.k03 Maturity Level 0 Unaware and Ad Hoc, key enterprise action | OWASP | I41 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
owasp.k07 Maturity Level 1 Experimentation without Guardrails, key enterprise action | OWASP | I21 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
owasp.k09 Maturity Level 1 Experimentation without Guardrails, key enterprise action | OWASP | I41 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
owasp.k12 Maturity Level 2 Policy-Defined, Human-in-the-Loop, key enterprise action | OWASP | I38 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
owasp.k18 Maturity Level 3 Integrated, Continuous Oversight, key enterprise action | OWASP | I39 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
mcp.01 Authorization (draft) | MCP authorization | S3 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
mcp.02 Authorization (draft) | MCP authorization | S3 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
aagmm.GD-03 GD-03 | AAGMM | I41 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
aagmm.GD-05 GD-05 | AAGMM | I58 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
aagmm.GD-06 GD-06 | AAGMM | I38 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
aagmm.GD-08 GD-08 | AAGMM | I41 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
ms.200.2.1 200: Repeatable, Business strategy | Microsoft | I08 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
ms.200.4.2 200: Repeatable, Technology and data | Microsoft | I10 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
ms.300.2.3 300: Defined, Business strategy | Microsoft | I21 (partly), I38 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
ms.400.2.1 400: Capable, Business strategy | Microsoft | I10 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
ms.400.2.3 400: Capable, Business strategy | Microsoft | S1 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.GOVERN-5.1 GOVERN 5.1 | NIST AI RMF | I57 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.GOVERN-5.2 GOVERN 5.2 | NIST AI RMF | I57 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.GOVERN-6.1 GOVERN 6.1 | NIST AI RMF | I11 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.MANAGE-2.4 MANAGE 2.4 | NIST AI RMF | I39 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
nist.MANAGE-4.3 MANAGE 4.3 | NIST AI RMF | I54 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
nist.MAP-1.1 MAP 1.1 | NIST AI RMF | I20 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.MAP-1.3 MAP 1.3 | NIST AI RMF | I20 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.MAP-1.4 MAP 1.4 | NIST AI RMF | S1 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.MAP-2.1 MAP 2.1 | NIST AI RMF | I20 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
nist.MAP-2.2 MAP 2.2 | NIST AI RMF | I22 (evidenced), I21 (partly) | evidenced · π Moderate · S1·R5·A1 ×3 |
nist.MAP-3.3 MAP 3.3 | NIST AI RMF | I20 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
nist.MAP-3.5 MAP 3.5 | NIST AI RMF | I38 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.MAP-4.1 MAP 4.1 | NIST AI RMF | I11 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.MEASURE-2.5 MEASURE 2.5 | NIST AI RMF | I22 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
nist.MEASURE-3.3 MEASURE 3.3 | NIST AI RMF | I57 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
aiact.art-12 Article 12 | EU AI Act | I41 (evidenced) | evidenced · π Moderate · S1·R5·A1 ×3 |
aiact.art-13 Article 13 | EU AI Act | I20 (partly), I22 (partly), S3 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
aiact.art-14 Article 14 | EU AI Act | I38 (partly), I39 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
aiact.art-26 Article 26 | EU AI Act | I54 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
aiact.art-73 Article 73 | EU AI Act | I54 (partly) | partly · π Moderate · S1·R5·A1 ×3 |
Upper layers
The public card is the front door. The same evidence, read with the governance rules, is this agent's GAMMA-Ray rating: freedom, governance, the level held, the grant and a work order. That layer is written for an operator.
The vault
Every source behind this card was captured and hashed into a public sgit vault (wcvaw1bi): captures.json lists each URL with the SHA-256 of what was fetched and when. Read key:
sgit_public_read_5ab82ed8d04719fb9fa7ffa28ef3cbd9eec71b4b03b2958ee62f470c3c1e6ba9:wcvaw1bi
With sgit: sgit clone sgit_public_read_5ab82ed8d04719fb9fa7ffa28ef3cbd9eec71b4b03b2958ee62f470c3c1e6ba9:wcvaw1bi.