> ## Documentation Index
> Fetch the complete documentation index at: https://docs.molesignal.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Mole Agent

> Investigate observability issues with Mole Agent, then control automation and write operations through approvals.

**Mole Agent** is the embedded operations agent and its product workspace.

Mole Agent can query logs, metrics, traces, RUM, profiles, alerts, on-call data, and reports through
server-registered tools. Mole Agent uses the signed-in user's organization and permissions; model-provided
organization identifiers are ignored.

<Warning>
  Mole Agent requires the `agent` license feature. Permissions still apply after the
  feature is enabled.
</Warning>

## Chat

Open **Mole Agent → Chat**, choose an available model, set the time range, and describe the
symptom. Pin a service, alert, environment, or stream as investigation context.

The response can include:

* a plain-language conclusion;
* tool calls and evidence summaries;
* links to source logs, metrics, traces, or incidents;
* token and model-cost metadata;
* a prompt template version and content hash for reproducibility.

Large raw tool results spill to object storage. Chat deletion first performs a best-effort
transcript archive.

## Investigations

Turn a complex question into a tracked investigation. An investigation records:

* status and current step;
* evidence and the resulting effect on the conclusion;
* competing hypotheses and confidence;
* related chat, approvals, and executions;
* the final summary and recovery verification.

## Automations

Automations save repeatable Mole Agent workflows. Configure triggers, allowed tools, steps,
approval behavior, and failure handling. Use **Dry run** to validate the workflow without real
writes.

## Approvals and executions

Write-capable tools never bypass the server execution policy.

| Risk | Default execution mode                                            |
| ---- | ----------------------------------------------------------------- |
| L0   | Automatic.                                                        |
| L1   | User confirmation.                                                |
| L2   | Single approval.                                                  |
| L3   | Dual approval.                                                    |
| L4   | Disabled; destructive remote tools have at least this risk floor. |

An approval stores the proposed operation and expires after the policy window. Before execution,
the server revalidates permission, resource state, parameters, risk, and idempotency. The execution
record keeps outcomes and recovery verification.

## Agent settings

Administrators can manage:

* agent profiles, allowed tools, timeouts, data scope, and network policy;
* model providers and write-only API keys;
* built-in, organization, and personal prompt overrides;
* tool registry policy and environment-specific risk floors;
* constrained MCP servers and synchronized tools;
* approval rules for each risk level.

Credentials remain encrypted and are never returned by the API.

## Permissions

* `agent.use` opens chat, investigations, automations, and execution history.
* `agent.manage` configures providers, prompts, tools, profiles, and workflows.
* `agent.approve` reviews controlled operations.

<CardGroup cols={2}>
  <Card title="Inbound MCP" icon="plug" href="/en-US/inbound-mcp">
    Connect Codex and other external AI clients to the governed runtime.
  </Card>

  <Card title="Tool catalog" icon="wrench" href="/en-US/tools">
    Review discovery, IAM, risk policy, approvals, and execution behavior.
  </Card>

  <Card title="Alerting" icon="bell" href="/en-US/alerting">
    Start an investigation from an active incident.
  </Card>

  <Card title="Enterprise Edition" icon="key" href="/en-US/editions/enterprise">
    Review the Agent entitlement and Enterprise security model.
  </Card>
</CardGroup>