Primary project / AI product system

Plato

Plato is an AI-assisted product workbench for turning ambiguous product work into durable plans, documents, and verifiable execution loops.

138k+ lines of code 241k+ with docs long-running iteration
  • AI workflow
  • Product systems
  • Python
  • TypeScript
  • Documentation

User value

Turning AI from a chat box into a durable product work loop.

Reduce task ambiguity

Turn rough product intent into clear goals, task frames, document structures, and reviewable next actions.

Preserve work continuity

Keep decisions, context, artifacts, and execution state available across long-running AI-assisted work.

Make output verifiable

Move AI work away from one-off generation and toward evidence, acceptance gates, recovery paths, and usable artifacts.

Product loop

Designed around repeatable AI-assisted execution.

The product idea is not to make AI answer faster. It is to make work easier to define, continue, inspect, and recover when a task spans many decisions and artifacts.

  1. 01 Capture intent
  2. 02 Build task context
  3. 03 Plan execution loop
  4. 04 Create project assets
  5. 05 Review evidence
  6. 06 Recover and continue

Product highlights

What this project proves.

Plato is evidence of product judgment: how to define Agent workflows, manage complexity, and turn generative work into operational value.

Use context governance to maintain stable task state instead of relying on temporary prompt history.

Make documents the center of work so plans, notes, and decisions become durable project assets.

Use human review gates before Agent output becomes project evidence.

Design the execution loop around recovery, continuation, and repeatable product operations.

Show sustained engineering capacity beyond prototype speed.

Connect writing and project proof into a complete story for AI product roles.

Concept model

Explain the product model first, then show the interface.

The concept, flow, and architecture diagrams make Plato easier to understand before visitors inspect screenshots. They frame the project as an Agent product system rather than a set of UI images.

Concept diagram explaining what an Agent application is
Agent product concept An Agent application combines task intent, context, tools, execution, evidence, and human control.
Plato main workflow concept diagram
Plato main flow Shows how user intent moves through planning, execution, review, artifact creation, and continued work.
Plato architecture concept diagram
System architecture Connects the product surface, execution runtime, context governance, file evidence, and review boundaries.

Product screenshots

Real work surfaces behind the product claim.

These screenshots focus on the behaviors that matter in Agent product work: clarification, session continuity, file evidence, diff review, and trust management.

Plato asks the user a question during execution
Runtime clarification Execution can pause and ask for user guidance instead of guessing through ambiguous decisions.
Plato session activity timeline
Session activity Activity records make the execution process inspectable and help users understand what happened in a session.
Plato changed files list
Changed files File-level evidence helps users review what the Agent changed and binds implementation output to concrete artifacts.
Plato file diff viewer
Diff review A diff-oriented review surface turns generated code into inspectable change evidence before acceptance.
Plato trust plane interface
Trust plane The trust plane makes permissions, risks, review, and acceptance visible instead of hiding them in chat history.

Scale evidence

Enough scale to show sustained product and engineering investment.

Updated Jun 12, 2026

Scope Files / dirs Lines
Python 代码:src 195 52,303
Python 代码:tests 100 33,517
前端代码:非测试 156 36,133
前端代码:测试 67 15,902
项目文档 313 102,884
代码合计 518 137,855
文档与代码合计 831 240,739

Role relevance

Why this matters for Agent product roles.

Agent products fail easily when they are treated as page design or prompt assembly. Plato focuses on a deeper product surface: state models, context contracts, recovery behavior, review gates, and artifact quality.

That makes it a strong recruiting case: it connects user confusion, product workflow design, technical constraints, and deliverable evidence into one system.

Related articles

Writing that explains the product thinking behind Plato.