Exemplar Dev
Welcome to the Exemplar Dev documentation.
- Website: exemplar.dev
- Console: console.exemplar.dev
- Pricing: exemplar.dev/pricing
Exemplar Platform
Exemplar Platform is the control plane for harness engineering in Agentic DevOps. It has two modules you can buy together (Full Platform) or separately:
- Harness — AI operations control plane: gateway, prompts, skills, memory, context, evals, and guardrails—with SDK and CLI surfaces for frameworks and terminal workflows. Hub: exemplar.dev/harness .
- Tools & MCP — Governed MCP tools from connectors—one endpoint for agent frameworks and MCP clients (Cursor, Claude, and more).
- Automation Studio — No-code / low-code agent orchestrator: build, trace, and approve workflows on a graph canvas in the console.
- Relay — See, control, and prove what Cursor and Claude coding agents do: bash, path, MCP, and prompt controls before side effects, with audit replay. Hub: exemplar.dev/relay .
Platform overview
-
Describe intent — Engineers or AI agents state what they need from the IDE; the platform resolves context via live graph query.
-
Policy gate — The policy engine evaluates intent against integrated sources (AWS, GitHub, Snyk, Jira, Notion, Linear, Confluence) and approves or blocks before any change runs.
-
Execute and audit — Approved actions run as controlled operations (restart, scale, rotate secrets, trigger pipelines) with a full record of who, what, when, and evidence—the same rules for humans and AI.
-
Context lake — AWS, GitHub, Snyk, Jira, Notion, Linear, and Confluence data converge in one graph so operations and agents share current service context.
-
Automation and governance — DevX Assist actions (restart, scale, rotate secrets, trigger pipelines) run through policy, approvals, and an audit log—not ad hoc shell access.
Exemplar is a unified developer platform for harness engineering, coding-agent governance, and AI-assisted DevX Assist workflows—governed end to end by Harness and Relay.
Core areas include:
- Harness — Gateway, prompts, skills, memory, context, evals, guardrails, plus SDK and CLI for production agents on any framework.
- Tools & MCP — Out-of-box connector tools on a unified MCP endpoint—efficient, governed, and usable from frameworks or MCP clients.
- Automation Studio — Visual agent orchestrator: triggers, logic, connectors, agents, human approval, live trace, and publish.
- Relay — Bash, path, MCP, and prompt controls for Cursor and Claude coding agents, with seats, policy, and audit replay.
- DevX Assist — AI-assisted operational workflows, Copilot / MCP in the IDE, and planned self-service production actions.
- Team and organization management — Organization members, invitations, teams, API tokens, and account-level settings.
AI Copilot for DevX Assist extends these workflows into editors and AI clients via MCP.
For the latest product updates, visit exemplar.dev .
Blog
Editorial posts on reliability engineering, platform operations, and AI-assisted workflows:
- Developer autonomy and the work that repeats after ship — Why post-launch operational work dominates software delivery.
- Agents, context, and guardrails on a unified platform — AI copilots, service context, governance, and operational automation.
- Moving from prompt and context engineering toward harness engineering — Prompt, context, and harness layers for production agents.
Next steps
- Harness — AI operations control plane (gateway, prompts, skills, memory, SDK, CLI).
- Tools & MCP — Governed connector tools for frameworks and MCP clients.
- Automation Studio — No-code / low-code agent orchestrator in the console.
- Relay — Coding-agent governance for Cursor and Claude.
- DevX Assist — AI-assisted operational workflows and production operations.
- Guides — Hands-on guides, including Loop Engineering on Exemplar.
- Account Settings — Organization management, teams, invitations, and API access.