Valdr Skills
Valdr Skills
{"content":"The docs in this section cover the five operator wrappers shipped in `valdr-packs/skills`. The same package also includes `valdr-workflow-tools` for authoring executable user tools.\n\nSkills let your coding assistant use the agents and instructions you have installed in Valdr. Choose a skill for the work you want to do, then name the agent or task in your request.\n\n\u003e [!IMPORTANT]\n\u003e **PM operator skills require MCP access.**\n\u003e Use **Vanguard** or **Sovereign**, according to the tools the loaded agent needs. Writing and validating local tool source with `valdr-workflow-tools` uses the pack CLI; executing installed user tools requires **Sovereign**.\n\n\u003e [!NOTE]\n\u003e Valdr ships **five wrappers**: `valdr-orchestrator`, `valdr-planner`, `valdr-executor`, `valdr-reviewer`, and `valdr-auditor`.\n\n## Who Skills Are For\n\nSkills are for developers who prefer working from the terminal but want Valdr's structure and traceability.\n\n- **Claude Code** or **Codex CLI** with MCP support\n- **Vanguard** or **Sovereign** subscription\n- **Valdr MCP server** running locally\n\n## Available Skills\n\n{{\u003c cards \u003e}}\n {{\u003c card link=\"valdr-orchestrator/\" title=\"Orchestrator\" icon=\"search\" subtitle=\"Navigate and discover your workspace\" \u003e}}\n {{\u003c card link=\"valdr-planner/\" title=\"Planner\" icon=\"clipboard-list\" subtitle=\"Structure ideas into executable plans\" \u003e}}\n {{\u003c card link=\"valdr-executor/\" title=\"Executor\" icon=\"play\" subtitle=\"Execute tasks with consistent workflow\" \u003e}}\n {{\u003c card link=\"valdr-reviewer/\" title=\"Reviewer\" icon=\"check\" subtitle=\"Quality gates that can't be skipped\" \u003e}}\n {{\u003c card link=\"valdr-auditor/\" title=\"Auditor\" icon=\"eye\" subtitle=\"Evaluate agent work with structured scoring\" \u003e}}\n {{\u003c card link=\"/valdr/docs/workflows/user-tools/quickstart/\" title=\"Workflow Tools\" icon=\"code\" subtitle=\"Build typed scripts and CLI adapters for agents and workflows\" \u003e}}\n{{\u003c /cards \u003e}}\n\n## What Ships Today\n\n| Skill | Use it to | Include in your request |\n|-------|-----------|-------------------------|\n| **Orchestrator** | Find work, inspect status, or coordinate a workflow | The project, sprint, or run you want to inspect |\n| **Planner** | Turn requirements into a plan and tasks | The goal, constraints, and intended outcome |\n| **Executor** | Work on an assigned task | The task key |\n| **Reviewer** | Review a task's implementation and evidence | The task key and intended reviewer |\n| **Auditor** | Evaluate an agent session | The session or task and intended auditor |\n\n## How Skills Work\n\nWhen you invoke an operator skill, it:\n\n1. Finds the agent and instructions for your request.\n2. Loads those instructions from your local Valdr workspace.\n3. Follows them to carry out the work.\n\nTo change how work is scored, reviewed, or assigned, update the corresponding prompt or capability in Valdr.\n\n## Before making changes {#shared-protocol}\n\nSkills inspect current records before updating them. Include a concrete project, task, or session in your request so the assistant can confirm it is acting on the right work.\n\nIf a response is lost, ask the assistant to check the existing operation before starting it again. For agents making MCP calls directly, follow the [request ID and retry guidance](/valdr/docs/valdr-mcp/#idempotency-and-client-request-ids).\n\n## Invoke the Current Set\n\nInvoke a wrapper in Claude Code or Codex CLI:\n\n```\n/valdr-orchestrator\n/valdr-planner\n/valdr-executor\n/valdr-reviewer\n/valdr-auditor\n```\n\n## Prompt Gallery\n\nStart with the wrapper slash command, then give it a concrete request:\n\n{{\u003c prompts \u003e}}\nUse valdr-orchestrator with verdandi to inspect this blocked workflow run and explain the cause.\nUse valdr-planner with freya to create a plan for adding audit-log export to the admin UI.\nUse valdr-executor for task WEB-142 and stop at the review gate when the work is ready.\nUse valdr-reviewer with sigrid for task WEB-142 and focus on risky shell commands.\nUse valdr-auditor with tyr-v2 to audit the latest executor session for WEB-142 and record a score run.\nUse valdr-workflow-tools to build a read-only CLI adapter with typed inputs and outputs, validate the pack, and show me the archive before import.\n{{\u003c /prompts \u003e}}\n\nFor handle-driven wrappers, naming the agent directly in your prompt is the fastest path. If you omit the handle, Valdr lists the matching registered agents and asks which one should take the request.\n\n## Installation\n\nRun the relevant Make target from a checkout of the `valdr-packs` repository. These targets install skill files into your coding assistant; they do not import a `.valdr-pack.tar.gz` archive into Valdr:\n\n```bash\nmake sync-skills-codex # User-wide: ~/.agents/skills/\nmake sync-skills-claude # User-wide: ~/.claude/skills/\nmake sync-skills # Project-local assistant directories in this checkout\n```\n\n`make sync-skills` installs the current wrappers for project-level Claude, Codex, and Gemini setups.\n\n## Next Steps\n\n- **[Orchestrator](valdr-orchestrator/)** — Route work through Gunnar, Mimir, Nikol, Skadi, Verdandi, or another registered orchestrator\n- **[Planner](valdr-planner/)** — Use when work needs a structured plan or task generation\n- **[Executor](valdr-executor/)** — Use when a task key exists and the work should begin\n- **[Reviewer](valdr-reviewer/)** — Use when a task needs review coverage or a verification gate\n- **[Auditor](valdr-auditor/)** — Use when you need a scored evaluation of a session","description":"Use the Valdr operator wrappers and user-tool authoring skill from your coding assistant.","title":"Valdr Skills"}
The docs in this section cover the five operator wrappers shipped in valdr-packs/skills. The same package also includes valdr-workflow-tools for authoring executable user tools.
Skills let your coding assistant use the agents and instructions you have installed in Valdr. Choose a skill for the work you want to do, then name the agent or task in your request.
Important
PM operator skills require MCP access.
Use Vanguard or Sovereign, according to the tools the loaded agent needs. Writing and validating local tool source with valdr-workflow-tools uses the pack CLI; executing installed user tools requires Sovereign.
Note
Valdr ships five wrappers: valdr-orchestrator, valdr-planner, valdr-executor, valdr-reviewer, and valdr-auditor.
Who Skills Are For
Skills are for developers who prefer working from the terminal but want Valdr’s structure and traceability.
- Claude Code or Codex CLI with MCP support
- Vanguard or Sovereign subscription
- Valdr MCP server running locally
Available Skills
Navigate and discover your workspace
Structure ideas into executable plans
Execute tasks with consistent workflow
Quality gates that can’t be skipped
Evaluate agent work with structured scoring
Build typed scripts and CLI adapters for agents and workflows
What Ships Today
| Skill | Use it to | Include in your request |
|---|---|---|
| Orchestrator | Find work, inspect status, or coordinate a workflow | The project, sprint, or run you want to inspect |
| Planner | Turn requirements into a plan and tasks | The goal, constraints, and intended outcome |
| Executor | Work on an assigned task | The task key |
| Reviewer | Review a task’s implementation and evidence | The task key and intended reviewer |
| Auditor | Evaluate an agent session | The session or task and intended auditor |
How Skills Work
When you invoke an operator skill, it:
- Finds the agent and instructions for your request.
- Loads those instructions from your local Valdr workspace.
- Follows them to carry out the work.
To change how work is scored, reviewed, or assigned, update the corresponding prompt or capability in Valdr.
Before making changes
Skills inspect current records before updating them. Include a concrete project, task, or session in your request so the assistant can confirm it is acting on the right work.
If a response is lost, ask the assistant to check the existing operation before starting it again. For agents making MCP calls directly, follow the request ID and retry guidance.
Invoke the Current Set
Invoke a wrapper in Claude Code or Codex CLI:
/valdr-orchestrator
/valdr-planner
/valdr-executor
/valdr-reviewer
/valdr-auditorPrompt Gallery
Start with the wrapper slash command, then give it a concrete request:
- Use valdr-orchestrator with verdandi to inspect this blocked workflow run and explain the cause.
- Use valdr-planner with freya to create a plan for adding audit-log export to the admin UI.
- Use valdr-executor for task WEB-142 and stop at the review gate when the work is ready.
- Use valdr-reviewer with sigrid for task WEB-142 and focus on risky shell commands.
- Use valdr-auditor with tyr-v2 to audit the latest executor session for WEB-142 and record a score run.
- Use valdr-workflow-tools to build a read-only CLI adapter with typed inputs and outputs, validate the pack, and show me the archive before import.
For handle-driven wrappers, naming the agent directly in your prompt is the fastest path. If you omit the handle, Valdr lists the matching registered agents and asks which one should take the request.
Installation
Run the relevant Make target from a checkout of the valdr-packs repository. These targets install skill files into your coding assistant; they do not import a .valdr-pack.tar.gz archive into Valdr:
make sync-skills-codex # User-wide: ~/.agents/skills/
make sync-skills-claude # User-wide: ~/.claude/skills/
make sync-skills # Project-local assistant directories in this checkoutmake sync-skills installs the current wrappers for project-level Claude, Codex, and Gemini setups.
Next Steps
- Orchestrator — Route work through Gunnar, Mimir, Nikol, Skadi, Verdandi, or another registered orchestrator
- Planner — Use when work needs a structured plan or task generation
- Executor — Use when a task key exists and the work should begin
- Reviewer — Use when a task needs review coverage or a verification gate
- Auditor — Use when you need a scored evaluation of a session