Agent Governance Positioning

Workflow Skill Router addresses one layer of agent sprawl: skill selection sprawl.

It does not replace scope contracts, runtime permissions, approval policies, or tool access controls. Instead, it gives agents a small, reviewable routing decision before work starts.

What problem it handles

As an agent environment grows, the number of available skills, tools, connectors, and workflows grows with it. A broad task can then trigger too many related instructions at once:

frontend, ui, browser, playwright, qa, design-system, github, docs, deployment

Workflow Skill Router narrows that list before execution. It asks the agent to choose:

• one Primary skill,
• only the Supporting skills that reduce risk or add required context,
• the skills that look related but should stay out,
• a short reason for the route.

What it does not handle

Workflow Skill Router is not a security boundary.

Use your existing governance layers for:

• scope contracts,
• runtime permissions,
• approval policies,
• tool access controls,
• sandboxing,
• secret handling,
• external system authorization.

The router assumes a skill is already available and allowed. It only decides whether that skill is useful for the current task.

Why this layer matters

Skill selection is small enough to review before the agent edits files or calls tools. That makes it useful as an early checkpoint:

Route: Frontend / Debugging > Browser reproduction > Single-page app
Use SKILL: vue-expert, systematic-debugging, playwright
Reason: vue-expert handles component behavior; systematic-debugging keeps the investigation causal; playwright captures the regression.

The user can correct this decision before execution begins.

Recommended governance stack

Workflow Skill Router fits best as a pre-execution layer:

Task request
  -> Scope contract
  -> Skill selection route
  -> Runtime permissions
  -> Tool approvals
  -> Execution and verification

This keeps the project honest: it controls instruction selection, not authorization.