Skip to content
@xiaojiou176-open

xiaojiou176-open

Public showroom for readable, reviewable, evidence-backed AI products.

xiaojiou176-open showroom hero

A public showroom for AI products built from real, underestimated workflow needs.
Start with four flagship doors, then explore how reading, decision, execution, runtime, and proof products grow out of the same product thesis.

Author profileEnter the front rowBrowse by jobWalk the hallsUse the showroom

Flagship doors Product halls Visible atlas Built in public

This page is a showroom, not a repo wall.
Start with one flagship door. If it clicks, the rest of the atlas will make sense.
Built in public out of Seattle, WA (PT).

This is not a shelf of demos, wrappers, or side projects. It is a public map of products that started from real workflow pain, then grew into systems for reading, deciding, executing, connecting, and proving work under real constraints.

1. 🚪 Enter through the Front Row

These are the four fastest entrances into the portfolio. They keep the canonical order, but they are not a prize ranking or a toy-project shelf.

Readable outputs from raw source streams.

  • Need: too much source material, but not enough clarity
  • What it is: a reading product that turns messy source streams into traceable documents, briefings, and grounded answers
  • Why it matters: this portfolio starts by helping people understand the world, not just collect more feeds

Choose under real constraints without crossing the line.

  • Need: school work is fragmented, but the decision still has real stakes
  • What it is: an academic decision workspace that unifies Canvas, Gradescope, EdStem, and MyUW
  • Why it matters: some decisions are high-stakes and boundary-sensitive, not toy demos

Make execution trustworthy, not just impressive.

  • Need: long-running AI work breaks when teams rely on chats, scripts, and after-the-fact logs
  • What it is: an open command tower for planning, delegation, tracking, replay, and proof
  • Why it matters: serious AI systems need inspection, governance, and recovery, not just polish

Give the rest of the universe one runtime to stand on.

  • Need: visible AI tools keep rebuilding the same access and routing layer from scratch
  • What it is: a shared provider runtime kernel for BYOK, Web/Login access, routing, and diagnostics
  • Why it matters: products above it stay honest only if somebody owns the runtime underneath

2. 🧭 Browse the Atlas by Job

If you care about... Start here What you get next
Readable outputs instead of feed overload SourceHarbor docsiphon extends that into corpus-building and documentation-grade extraction.
Choosing under pressure, constraints, or messy interfaces OpenCampus DealWatch carries the same decision discipline into a consumer shopping workflow.
Workflows that can be inspected instead of trusted on faith OpenVibeCoding prooftrail and ui-automation-control-plane push that further into proof, replay, and repair.
Runtime, access, proof, and recovery Switchyard prooftrail and agent-exporter show what that foundation enables once systems have to be inspected and archived.
Browser-native, user-facing AI products Shopflow DealWatch and multi-ai-sidepanel show how the same depth turns into products real users can feel quickly.

3. 🏛️ What Lives in Each Hall

Hall What pain it fixes What these repos actually do Representative repos
📚 Reading and Knowledge Products
  • too much source material
  • not enough clarity
  • rebuild feeds, documents, and source streams
  • turn them into reading-grade outputs people would actually keep
SourceHarbor, docsiphon
🎯 Real-World Decision Workspaces
  • the decision is real
  • the information is fragmented
  • the boundary matters
  • turn scattered academic or shopping surfaces into safer decision environments
  • replace noisy dashboards with clearer judgment
OpenCampus, DealWatch
🧭 Control and Workflow Systems
  • a workflow may run
  • but nobody can explain, inspect, or recover it
  • add governed execution
  • add replay and proof to serious AI work
OpenVibeCoding
🧠 AI Workbenches
  • a chat box is too thin for real work
  • turn prompting, comparison, delivery, and archive work into real workspaces
  • avoid one-shot interfaces
multi-ai-sidepanel, provenote, openui-mcp-studio, agent-exporter
🛡️ Trust and Recovery Labs
  • risky actions need preview
  • rollback and forensic discipline matter
  • build recovery, snapshot, and verification tools
  • make trust come from inspectability
apple-notes-snapshot, apple-notes-forensics, movi-organizer
🔌 Runtime and Proof Foundations
  • every visible product breaks when runtime, proof, or recovery is weak
  • provide the shared runtime kernel
  • provide the evidence substrate that keeps the universe honest
Switchyard, prooftrail, ui-automation-control-plane
🛍️ Browser Product Families
  • system depth should feel useful without reading architecture docs
  • turn browser workflows into product families
  • avoid isolated utilities
Shopflow

4. ⭐ Flagship Reserve

If you want the strongest second-ring doors after the front row, open these next. This shelf should widen the picture, not repeat the same four doors again.

  • docsiphon — if you want the strongest extra doorway into CLI/corpus engineering and document-to-artifact pipelines.
  • Shopflow — if you want the clearest proof that this universe can ship consumer-facing browser product families.
  • DealWatch — if you want the cleanest browser-native example of compare-first, evidence-backed decision making.
  • agent-exporter — if you want the archive, governance, and transcript-to-evidence side of the system.

5. 🔍 How to Use This Showroom

  • Use the front row if you want the shortest path in.
  • Use the halls if you want the portfolio grouped by the job it solves.
  • Use the routes if you already know the problem you care about.
  • Use the atlas if you want the deeper layers to stay visible instead of disappearing behind four highlight repos.

Pinned Loading

  1. codeflow codeflow Public

    OpenVibeCoding is the command tower for AI engineering: plan, delegate, track, resume, and prove long-running work across Codex and Claude Code.

    Python 207 32

  2. webai-bridge webai-bridge Public

    Shared provider runtime for AI apps.

    TypeScript 8 1

  3. BrewMe BrewMe Public

    Reader-first source intelligence and MCP server for YouTube, Bilibili, RSSHub, and RSS, with grounded search, published reader docs, public CLI/TypeScript SDK surfaces, and plugin-grade builder bun…

    Python 2

  4. campus-copilot campus-copilot Public

    Academic decision workspace with cited study context from Canvas, Gradescope, EdStem, and MyUW.

    TypeScript 2

  5. shadcn-brief shadcn-brief Public

    Public repo for MCP-native UI delivery and review across Codex, Claude Code, OpenCode, and OpenClaw.

    TypeScript 1 1

  6. dealwatcher dealwatcher Public

    Open-source compare-first grocery price tracking with compare-aware watch groups, effective price, health, and alert history.

    Python 1

Repositories

Showing 10 of 16 repositories
  • campus-copilot Public

    Academic decision workspace with cited study context from Canvas, Gradescope, EdStem, and MyUW.

    xiaojiou176-open/campus-copilot's past year of commit activity
    TypeScript 2 MIT 0 0 6 Updated Sep 1, 2026
  • notebooklab Public

    Messy long context -> structured insight -> auditable markdown, research threads, and inspectable outcomes.

    xiaojiou176-open/notebooklab's past year of commit activity
    Python 1 MIT 0 0 21 Updated Aug 25, 2026
  • shopflow-suite Public

    Chrome-first shopping extension product family with 8 storefront apps, 1 Suite shell, a reviewer-facing release shelf, and a read-only stdio MCP surface.

    xiaojiou176-open/shopflow-suite's past year of commit activity
    TypeScript 1 MIT 0 0 14 Updated Aug 24, 2026
  • fileman Public

    Review-first local AI file organizer with dry-run apply, rollback, and agent-safe MCP.

    xiaojiou176-open/fileman's past year of commit activity
    Python 1 MIT 0 0 29 Updated Aug 24, 2026
  • dealwatcher Public

    Open-source compare-first grocery price tracking with compare-aware watch groups, effective price, health, and alert history.

    xiaojiou176-open/dealwatcher's past year of commit activity
    Python 1 MIT 0 0 14 Updated Aug 20, 2026
  • pagestress Public

    AI-native WebUI stress lab for localhost-first browser experiments, governed proof, and MCP-ready workflows.

    xiaojiou176-open/pagestress's past year of commit activity
    TypeScript 1 MIT 0 0 14 Updated Aug 12, 2026
  • notes-recover Public

    Copy-first Apple Notes recovery and review toolkit for macOS, with AI-assisted triage, evidence-backed case Q&A, and a read-mostly MCP server for Codex / Claude Code style local agent workflows.

    xiaojiou176-open/notes-recover's past year of commit activity
    Python 2 MIT 1 0 6 Updated Aug 4, 2026
  • agent-export Public

    Archive and governance workbench for AI agent transcripts, archive shells, and evidence routing.

    xiaojiou176-open/agent-export's past year of commit activity
    Rust 1 MIT 0 0 6 Updated Jul 27, 2026
  • codeflow Public

    OpenVibeCoding is the command tower for AI engineering: plan, delegate, track, resume, and prove long-running work across Codex and Claude Code.

    xiaojiou176-open/codeflow's past year of commit activity
    Python 207 MIT 32 0 31 Updated Jul 15, 2026
  • multi-ai-panel Public

    Compare-first, browser-native AI compare workspace for ChatGPT, Gemini, Perplexity, Qwen, and Grok, with governed coding-agent integrations for Codex and Claude Code browser workflows.

    xiaojiou176-open/multi-ai-panel's past year of commit activity
    TypeScript 2 MIT 0 0 16 Updated Jun 29, 2026