Bring Moona Knowledge
into the tools you
already use.
Give your AI tools access to Moona Knowledge's published Intelligence, Records, Agent Execution Risks and Protocol evidence, so they can reason with evidence-backed execution-authority knowledge.
Moona Knowledge is the knowledge and reasoning layer for AI-agent execution authority.
It helps an AI tool understand:
- what is known
- what evidence supports it
- which execution-authority risks apply
- which published protocol evidence is relevant
- what remains uncertain
Moona Knowledge does not execute, approve, block, or monitor agent actions.
It returns evidence-backed reasoning. Runtime interception, approval workflow and enforcement belong to Moona Product, not to this endpoint.
Start from a real question.
“A deployment was approved, but the code changed after approval and the agent still shipped it. What went wrong?”
- 1
Scenario
You describe what happened, in your own words.
- 2
Canonical risk
The relevant published Agent Execution Risks are retrieved by their real ids.
- 3
Evidence
Each finding stays attached to its canonical record, provenance and sources.
- 4
Protocol
Connected published protocol evidence is surfaced where it exists.
- 5
Uncertainty
What the evidence does not establish is stated, not filled in.
This is the shape of the reasoning, not a sample answer. Real ids, evidence and wording come from the live endpoint when you ask it yourself.
Connect Moona Knowledge MCP
Three steps. No Moona account.
Choose your MCP client
Pick the MCP-compatible tool you want to connect.
Verified means Moona tested this wiring against the live endpoint. Host dependent means the connector is documented and tested manually, but availability depends on your own account, plan or workspace settings. Other MCP clients are not verified by Moona.
Add Moona Knowledge
Public endpoint. No Moona Knowledge account or API key required.
$ claude mcp add --transport http moona-knowledge https://moona.ozlunara.com/mcp
Your AI host may require its own account and may restrict custom MCP connections depending on its plan or workspace settings.
Ask your first question
Start with one of the scenarios we use to verify Moona Knowledge.
A deployment was approved, but the code changed after approval and the agent still shipped it. What went wrong?An agent is about to perform a consequential production action, but the available context doesn't show whether an applicable approval or policy exists. Is it authorized?A planning agent delegates a task to a specialist sub-agent using its own credentials. What authority questions need to be resolved before it acts?Ask normally. You don't need to know Moona's terminology.
Six things to ask it for.
Each one maps to a capability in the current Moona Knowledge implementation.
Identify the canonical execution-authority risks that apply to a described scenario, with their real registry ids.
Reason about changes between what was approved and what is about to run: action, artifact, target, actor, environment, constraints, delegation and temporal scope, where the current authority-resolution model supports them.
Retrieve canonical Moona Knowledge objects by id or search, with their provenance, sources and connected evidence.
Surface published protocol evidence connected to an execution-authority question, using the relationships already recorded in the canonical registry.
Reason about authority that passes from a parent agent to a specialist agent, including credentials and the actual execution context.
Preserve unknown or missing authority instead of silently treating technical capability or absent evidence as authorization.
What changes when Moona Knowledge is called.
On its own
A general AI model reasons from its training and from whatever is in your conversation.
With Moona Knowledge connected
It can additionally retrieve Moona's published canonical knowledge:
- Intelligence and Records with their provenance and sources
- Agent Execution Risks by their real registry ids
- Published protocol evidence connected to those risks
- The recorded relationships between all of them
This adds knowledge. It does not guarantee a better answer, and no benchmark comparison is claimed here.
Missing authority is not authorization.
“An agent is about to perform a consequential production action, but the available context doesn't show whether an applicable approval or policy exists. Is it authorized?”
Moona Knowledge's authority-resolution model distinguishes evidenced, inferred and unknown context, and keeps the gap visible instead of converting silence into a yes. Its governance outputs are:
Moona Knowledge provides decision intelligence and learning governance. It may produce an evidence-backed analysis or recommendation for a scenario, but it does not itself intercept, execute, or enforce a real production action, and must never be described as doing so. Runtime interception, approval workflow, and enforcement belong to Moona Product (or another evidenced enforcement system). Enforcement outcomes can flow back into Moona Knowledge as execution evidence, but a Moona Knowledge answer is never proof that enforcement occurred.
Everything it returns resolves to a published page.
These are live entries from the canonical corpus this endpoint reads, rendered from the same data an MCP client receives.
Slack Put the Approval Where the Conversation Is. It Did Not Put the Enforcement There.
On 20 August 2026 Slack, a Salesforce company, launched Slack Code: a new channel type that partner coding agents create through a Slack API, carrying a plan, a repository and branch, code diffs and a live preview. Slack documents that anyone in the channel can pause, redirect or stop an agent mid task, that agents inherit Slack's permissions and admin controls so no new identities are provisioned, and that the channel archives itself into an audit log when the task ends. Slack also says that for high stakes moves, like pushing code to production, the agent packages its work for an expert to sign off on, right in the channel. That is a description of a workflow, not of an enforcement mechanism. Slack states plainly that Slack Code is not a coding model, a harness or an agent runtime, which means the code does not execute in Slack and neither does the deployment. Nothing published establishes who counts as an expert, whether a sign off binds to a specific diff, commit or deployment, whether an agent can proceed without one, or what downstream system enforces the decision. Moona Intelligence reads the launch as the approval moment moving onto a general purpose chat surface while the enforcement point stays somewhere Slack does not document.
Read the canonical recordApproval not bound to the executed action
What was approved and what executed diverge, because part of what an action does is decided by state the approval step never inspected, or the approval attaches to nothing in particular.
Connected to 13 published records and 9 protocols in the canonical registry.
See the canonical riskEP Authorization Receipts (EMILIA Protocol)
Implementations MUST reject an approval request whose action hash does not match a locally recomputed hash of the presented Action Object
EMILIA's requirement that an approval be rejected unless the action hash matches a locally recomputed hash of the exact action object is the binding these cases lack, where an approved command's behaviour is decided by state the approval never inspected.
See the canonical protocolWorks across the MCP ecosystem.
Verified with selected MCP clients and libraries.
Compatibility depends on the MCP capabilities supported by each client. “Verified” means Moona has tested a real connection and tool call against Moona Knowledge MCP. It does not imply partnership, endorsement or official certification by any of the projects named above.
Test Moona Knowledge more deeply
Once you are connected, these are the ten areas Moona works through when verifying a host end to end. Run them against your own work, in your own tool.
- Connection to the endpoint from your host
- Explicit retrieval by canonical id
- Natural-language retrieval
- Connected synthesis across Intelligence, Records, Risks and Protocols
- Provenance: every cited id and url resolves to a real published page
- Unknown authority and gap filling
- Delegated authority
- Ambient discovery inside your host's normal workflow
- Cross-host reliability
- Usefulness on your own real work
Bring Moona Knowledge into your workflow.
Use Moona's published Intelligence, Records, Risks and Protocols without leaving your MCP-compatible tool.