Vision
PointDev captures are currently consumed locally — one person captures, their local AI agent reads it. The structured output format is designed to go further: captures should be shareable between people, tools, and organizations.
A client opens a staging site, circles the hero text, says "this feels too corporate." The capture — voice + spatial annotation + DOM context + computed styles — arrives in the developer's project management tool, inbox, or directly into Claude Code. No Loom video to scrub through. No screenshot with a red arrow drawn in Preview. The capture carries full technical precision regardless of whether the person making it is technical.
What this enables
- Cross-team feedback: designer → developer, client → agency, QA → engineering
- Tool interoperability: captures produced by PointDev consumed by any tool that reads the format
- Asynchronous collaboration: captures are self-contained — no "which page were you on?"
Implementation layers (future)
- Export format — the JSON schema from the prompt format spec (M5) is the foundation. Captures need to be self-contained and portable.
- Import/viewer — a lightweight web viewer that renders a capture without requiring the extension. Enables recipients who don't have PointDev installed.
- Relay service — WebSocket or HTTP relay that routes captures from extension to any endpoint (email, Slack webhook, project management API, AI agent). Builds on the bridge server (M6).
Scope note
This issue captures the strategic direction. The format spec (M5) and bridge server (M6) fund the infrastructure that makes sharing possible. The relay service and web viewer are follow-up grant scope.
Related: #10 (output formats), #12 (bridge server), #35 (community engagement)
Vision
PointDev captures are currently consumed locally — one person captures, their local AI agent reads it. The structured output format is designed to go further: captures should be shareable between people, tools, and organizations.
A client opens a staging site, circles the hero text, says "this feels too corporate." The capture — voice + spatial annotation + DOM context + computed styles — arrives in the developer's project management tool, inbox, or directly into Claude Code. No Loom video to scrub through. No screenshot with a red arrow drawn in Preview. The capture carries full technical precision regardless of whether the person making it is technical.
What this enables
Implementation layers (future)
Scope note
This issue captures the strategic direction. The format spec (M5) and bridge server (M6) fund the infrastructure that makes sharing possible. The relay service and web viewer are follow-up grant scope.
Related: #10 (output formats), #12 (bridge server), #35 (community engagement)