Capabilities

What Is a Capability?

A capability is a named, trackable aspect of a project's functionality. Romyq uses capabilities to understand what a project has, what it still needs, and whether the readiness score is accurate.

The capabilities Romyq tracks by default map to the readiness categories:

• Core Functionality — Core Features, Database, Validation
• Testing — Testing
• Security — Security, Authentication, Authorization
• Operations — Deployment, Observability, Documentation

Capability States

Each capability can be in one of three states:

Viewing capabilities

romyq capabilities              # list all tracked capabilities
romyq capabilities list         # same as above
romyq capabilities --json       # machine-readable output

Capabilities are also shown in the readiness output:

romyq readiness

mission readiness: 67 / 100  (Approaching)

  Core Functionality  45%    Core Features: partial, Database: complete
  Testing              0%    Testing: missing
  Security            83%    Security: complete, Authentication: partial
  Operations          33%    Deployment: missing

Updating capabilities manually

romyq capabilities set "Authentication" complete
romyq capabilities set "Testing" partial
romyq capabilities set "Deployment" missing

Valid status values: complete, partial, missing.

Inferring capabilities from task history

romyq capabilities infer

Romyq reads the task history and infers capability states based on which tasks have completed. Useful after manually editing files outside the loop.

How capabilities are tracked automatically

During planning, the planner tags each task with one or more capability labels. As tasks complete, their capability tags are aggregated into the project capability model stored in .romyq/project_state.json.

Capabilities and the recommendation engine

• 5 or more consecutive task failures trigger Review, regardless of capability state.
• All phases complete but done criteria not yet satisfied triggers Review.
• All phases complete and all done criteria satisfied triggers Stop.

The Stop recommendation does not require capabilities to be manually verified — it relies on phase and done-criteria completion.