Research
Research
Contributor research dossiers separated from implementation roadmap items.
Research
This section holds research dossiers: external landscape scans, feasibility notes, source ledgers, and brainstorm material. Research can inform roadmap items, but it is not itself an implementation commitment.
Roadmap pages should stay task-shaped: current status, intended implementation, acceptance criteria, and links to the research that justifies the plan. When a research page becomes actionable, promote the resulting task to /roadmap/ and keep the evidence here.
Current dossiers
| Area | Dossiers |
|---|---|
| Agent orchestration | Program research, Orca ADE, conversation capture, terminal observation, memory, multi-agent collaboration, workflow systems |
| Agent runtimes | Package-manager support |
| Agent telemetry | Token and cost telemetry |
| Authentication | Claude auth strategy |
| Isolation | smolvm backend, OrbStack isolated machines |
| Security | Isolation architecture, process sandboxing, sandbox backends, credential exposure |
| CI and testing | CI performance/speed, Rust tooling, visual snapshots |
| Documentation and infrastructure | Documentation repository strategy, code-intelligence tooling |
| Operator/product/desktop | Host-affordance preflight, public attribution, Desktop Agent Hub |
| Runtime | Devcontainer parity |
| Session capture | Session capture (SpecStory internals, agent session stores, market scan — evidence for Déjà Vu) |
| Backlog | Unscheduled platform ideas |
| Token optimization | Token-optimization techniques, Token-optimization tools, jackin Context Engine spec |