Secureagentics is built around a small set of concepts that work together to give you visibility and control over your AI agents in production. Understanding these building blocks will help you get the most out of the platform.Documentation Index
Fetch the complete documentation index at: https://docs.adrian.secureagentics.ai/llms.txt
Use this file to discover all available pages before exploring further.
Agents
Agents
An agent is any AI-powered process you register with Secureagentics — a chatbot, an autonomous workflow, a code assistant, or any other system that sends and receives data using a language model.Once registered, each agent gets a unique ID (e.g.,
agt_abc123) that links all of its activity back to a single, trackable entity in your dashboard.Example: You run a customer support chatbot powered by GPT-4o. You register it as an agent named support-bot so Secureagentics can track every prompt it sends and every response it receives.Events
Events
An event is a discrete action or activity reported by an agent. Events are the raw data that Secureagentics collects, analyzes, and displays. Common event types include:
Example: Every time your support bot sends a message to GPT-4o, your instrumentation fires a
| Type | Description |
|---|---|
prompt | A prompt sent to a language model |
completion | A response received from a language model |
tool_call | An external tool or function invoked by the agent |
error | An exception or failure encountered during execution |
prompt event. When the model responds, it fires a completion event. Both events appear in the agent’s timeline in your dashboard.Policies
Policies
A policy is a rule that Secureagentics evaluates against your agents’ events in real time. Policies let you define what your agents are and aren’t allowed to do — and what should happen when a rule is violated.Policies can inspect event content (e.g., detect sensitive data in a prompt), enforce rate limits (e.g., no more than 1,000 tokens per minute), or restrict which models an agent may call.When an event violates a policy, Secureagentics can block the action, flag it for review, or trigger an alert — depending on how you configure the policy’s enforcement mode.Example: You create a policy called
no-pii-in-prompts that blocks any prompt containing a credit card number or Social Security Number from being forwarded to the model. If your agent accidentally includes PII, the policy intercepts it before it leaves your environment.Alerts
Alerts
An alert is a notification triggered when something noteworthy happens — a policy violation, an unusual spike in activity, an agent going offline, or any condition you define.You can route alerts to email, Slack, PagerDuty, or a webhook endpoint. Each alert includes context about the triggering event so you can investigate quickly.Example: You configure an alert that fires whenever your
data-pipeline-agent makes more than 50 tool calls in a single minute. When the threshold is crossed at 2 a.m., you receive a Slack message with a link to the offending event sequence.Audit Logs
Audit Logs
The audit log is an immutable, time-ordered record of everything that has happened across your agents and your Secureagentics account. This includes agent events, policy evaluations, configuration changes, and user actions in the dashboard.Audit logs are retained according to your plan and can be exported for compliance, incident response, or internal review.Example: After a security incident, your team needs to reconstruct exactly what your agent did in the 30 minutes before it was flagged. You export the audit log for that time window and share it with your security team — every prompt, completion, and tool call is accounted for.
Workspaces
Workspaces
A workspace is an isolated environment within your Secureagentics account. Each workspace has its own agents, policies, alerts, API keys, and audit logs, making it easy to separate concerns across teams, environments, or customers.A common pattern is to create separate workspaces for
production, staging, and development so that test traffic never mingles with real activity.Example: Your platform team manages AI agents for three internal products. They create a workspace for each product — workspace-search, workspace-support, and workspace-analytics — so each product team sees only their own agents and can manage their own policies independently.Ready to put these concepts into practice? Follow the Quickstart to register your first agent and send your first event in under 5 minutes.