Built-in orchestrators
Valdr’s built-in orchestrators are the agents you use when work needs structure instead of another ad-hoc chat. Each one owns a different control-plane job: Gunnar helps you understand and create PM work, Mimir keeps workspace knowledge searchable and useful, Nikol helps you design the agent system itself, and Skadi coordinates sprint execution.
This section covers the orchestrators in the Valdr Sovereign pack. The full pack includes more specialists, executors, reviewers, planners, and auditors, but these orchestrators are the command layer users reach for when they want predictable agent operations.
Note
Local-first: Valdr runs entirely on your machine. Agent definitions, prompts, capabilities, tasks, reviews, sessions, and history live in your local workspace and remain until you delete them.
Choose the right orchestrator
| Orchestrator | Handle | Use when you need to |
|---|---|---|
| Gunnar | gunnar | Navigate projects, tasks, sprints, agents, prompts, capabilities, and create structured tasks |
| Mimir | mimir | Curate knowledge sources, refresh indexes, search docs/code, map symbols, and capture agent memory |
| Nikol | nikol | Design agents, author capabilities, assemble manifests, and maintain Valdr packs |
| Skadi | skadi | Prepare sprints, assign executors, route reviews, launch sessions, and keep delivery moving |
How they work
The orchestrators stay lean at launch. Each starts with a focused system capability, then hot-loads deeper workflow guidance only when the request needs it. That keeps everyday navigation fast while still giving complex operations the detailed instructions they need.
Use them from the Orchestrator UI when you want a conversational command center, or launch them through your configured presets when another agent needs to delegate a workflow. Start with the smallest direct request that describes the outcome:
- “Gunnar, show me active tasks for this project and flag anything blocked.”
- “Mimir, find the source of truth for our release workflow and show the relevant code symbols.”
- “Nikol, help me design a docs reviewer agent for our pack.”
- “Skadi, prepare the next sprint from these tasks and check launch readiness.”
Good orchestrator use is explicit. Ask for the state you want inspected, the workflow you want prepared, or the handoff you want coordinated. The agent fetches current context first, then uses Valdr’s MCP tools and local PM state to make the next move inspectable.
Use orchestrators from the CLI
Power users can run the same orchestrators from Claude Code or Codex CLI through the valdr-orchestrator wrapper. Invoke the wrapper, name the orchestrator handle, and make the request direct:
/valdr-orchestrator
Use gunnar to show open tasks for project VALDR and summarize blockers./valdr-orchestrator
Use mimir to search project VALDR for session launch idempotency and show the source paths./valdr-orchestrator
Use nikol to inspect the pack and recommend where this reviewer workflow should live./valdr-orchestrator
Use skadi to prepare the next sprint from these tasks and report launch-readiness gaps.The wrapper resolves the handle, loads that orchestrator’s prompt, and then follows the loaded prompt as the source of truth. This keeps CLI sessions aligned with the current pack instead of relying on copied instructions.
Orchestrator docs
Start with Gunnar when you are unsure where work lives, Mimir when the answer depends on searchable workspace knowledge, Nikol when the agent system itself needs maintenance, and Skadi when execution needs coordination across tasks, reviewers, and sessions.
Need the specialists that own planning, review, and audit gates? Read Built-in workflow agents for Freya, Sigrid, and Tyr.