Romyq · Public Roadmap

Roadmap

This roadmap is honest about the current state of the project. Each section is clearly labeled. Features are not presented as available until they are available.

Implemented

v0.10.3

These features are available in the current release.

  • Lifecycle generation
    8-stage lifecycle from mission definition to verified completion. Persists across sessions in .romyq/.
  • Complexity profiles
    Three profiles: Basic (3 phases), Intermediate (5 phases), Advanced (7 phases). Selected at init.
  • Project memory
    Persistent state in .romyq/: task history, knowledge base, execution memory, rules, decisions, capabilities, lifecycle, and project constitution. Human-readable JSON and Markdown.
  • Governance engine
    Constitution-based rule enforcement. Proposed tasks evaluated before execution. Violations trigger replanning.
  • Readiness scoring
    Weighted aggregate of phase completion, capability coverage, done criteria, and governance state.
  • Recommendation engine
    Emits Continue, Pause, Review, or Stop after every task. Re-evaluated after every operator command.
  • Timeline
    Chronological, append-only log of every phase transition, task outcome, and operator command.
  • Operator shell
    Interactive REPL with built-in commands: status, roadmap, phase, readiness, recommendation, pause, resume, stop, rules, knowledge, dashboard. Free-text sends a steering instruction.
  • Provider transparency
    Planning and execution providers are configurable independently. DeepSeek, Gemini, Claude, OpenAI-compatible.
  • Multi-project support
    Multiple projects managed from a single session. Each maintains independent state, memory, and rules.
  • Decision log
    Every governance decision recorded in .romyq/decisions.json. Human-readable. Diffable.
  • Error recovery
    Lifecycle state preserved on failure. Resume from last checkpoint with romyq resume.

In Progress

active

Work underway in the current development cycle.

  • Documentation
    Expanding coverage of all commands, concepts, and lifecycle stages.
  • Test coverage
    Expanding test suite beyond the current 1,555 tests.

Planned

v0.11

Confirmed scope for the next minor release. Not yet implemented.

  • Bounded improvement cycles
    Allow multiple revision passes per task. Current implementation does a single pass then moves on.
  • Structured lifecycle reviews
    Formal review step at the boundary between phases. Currently phase transitions happen automatically.
  • Enhanced visibility tooling
    Better dashboard output and status reporting. The current dashboard is functional but limited in what it exposes.
  • Improved provider coordination
    Better handling of planning provider failures and fallback behavior.

Research

exploring

Being evaluated. No timeline. Not committed.

  • Checkpoint and rollback
    Ability to snapshot lifecycle state at arbitrary points and restore. Architecture not yet decided.
  • Better knowledge extraction
    More accurate extraction of project facts from large codebases. Current approach works but has precision limits.
  • Lifecycle templating
    Pre-defined lifecycle templates for common project types (REST API, CLI tool, library). Under consideration.

Potential

future

Ideas that may or may not ever happen. Not commitments.

  • Desktop UI
    A graphical interface for the dashboard. Low priority. The terminal interface is the intended primary interface.
  • Team workflows
    Multiple operators contributing to a single lifecycle. Significant architecture work required. Not currently planned.
  • Cloud-optional sync
    Optional sync of lifecycle state across machines. Explicitly not planned in the near term. Local-first is not a constraint we plan to relax.

Not Planned

out of scope

Features explicitly outside the scope of this project.

  • SaaS or hosted offering
    Romyq will not become a cloud service. This is a design principle, not a temporary constraint.
  • Telemetry or analytics
    No usage data will ever be collected. The project has no server-side infrastructure to receive it.
  • AI model fine-tuning
    Romyq orchestrates existing LLM providers. It does not train or fine-tune models.
  • IDE plugin
    Not planned. The CLI interface is the intended interface. IDE integration would add maintenance burden without clear benefit.

Questions about the roadmap? Open a GitHub Discussion ↗ or contact us.