Rye Roadmap

This roadmap is intentionally undated. Rye should advance by capability maturity and real adoption pressure, not fixed calendar promises.

Onboarding and Purpose

• Build an agent-driven onboarding flow around onboarding_scope that asks in human terms what function Rye is assisting, why it matters, what is out of scope, and what would signal that purpose has changed.
• Support piecemeal adoption by allowing multiple active scopes in one organization, each with distinct sources, plugins, policies, and autonomy gates.
• Add scope revision workflows for repeated context_gap candidates, source drift, new organizational goals, and policy changes.

Plugin System

• Move non-core vocabulary and behavior into loadable plugins with manifests, dependencies, supersession metadata, validators, admin contributions, and onboarding questions.
• Keep core Rye focused on graph primitives, assertions, events, provenance, artifacts, security, and convention helpers.
• Mature rye-org as a foundational plugin for people, systems, departments, mission, vision, goals, related organizations, policies, and procedures.
• Add plugin validation that compares manifests against active scope policies before agents insert or promote graph changes.

Source Intake and Evidence

• Make source, retrieval channel, and context first-class distinctions across all intake paths, including Composio, native connectors, direct APIs, exports, logs, and files.
• Preserve enough source evidence for replay, audit, and confidence without forcing every provider record to become a first-class graph node.
• Add source-specific intake profiles for Slack, email, meetings, files, tabular imports, logs, and application events.
• Expand post-intake review packets that show source inventory, pending confirmations, context gaps, candidate queues, retention exposure, and blocked promotions.

Storage and Retention

• Treat storage growth as a cross-cutting design constraint in every connector, plugin, and admin view.
• Add retention classes for evidence, context signals, low-signal material, raw replay artifacts, and noise summaries.
• Later, add pruning or garbage-collection jobs that preserve accepted evidence anchors while collapsing stale or low-value source material.

Admin Observability

• Focus the admin interface on Rye structure: graph exploration, source provenance, context gaps, candidate review, plugin policy, assertion history, disputes, and scope health.
• Keep domain record management in domain applications that understand Rye metadata, rather than making Rye admin the operational UI for every app.
• Allow plugins to contribute admin tabs, dashboard cards, policy views, candidate groupings, and validation explanations.

Agent Operation

• Provide agents with compiled scope policy bundles before collection, classification, evidence extraction, or promotion.
• Separate collector, classifier, evidence-extractor, promotion, dispute, and cleanup responsibilities.
• Add audit trails for why an agent considered evidence relevant, why it chose a node/edge/assertion type, and which policy allowed or blocked the action.
• Support agent-only handling only after a scope policy and repeated review outcomes make the action safe enough.

Graph Quality

• Strengthen candidate review, dispute handling, supersession, and confidence modeling for subjective knowledge extraction.
• Add quality reports for orphan nodes, unsupported edges, stale assertions, duplicate entities, repeated context gaps, and plugin-policy violations.
• Add migration paths for plugin supersession when a better plugin replaces an earlier good-enough convention.

Integration Surface

• Build repeatable Composio intake recipes that land source accounts, containers, source items, artifacts, context gaps, and pending confirmations without over-promoting business facts.
• Add native connector/plugin paths where direct integration provides better evidence, provenance, or control than a generic channel.
• Support indirect sources such as logs and audit trails as evidence sources with their own retention and confidence policies.

Security and Compliance

• Expand policy gates for sensitive sources, direct messages, private files, regulated data, and restricted teams.
• Add process-audit views that show who or what made each onboarding, classification, promotion, and scope-revision decision.
• Keep human-readable policy records so technical Rye admins can communicate behavior to organizational leaders and compliance reviewers.