|
2 | 2 |
|
3 | 3 | Welcome to the AI-OS wiki. |
4 | 4 |
|
| 5 | +The wiki is a curated browsing surface. The canonical source of truth remains in the repository docs. |
| 6 | + |
5 | 7 | ## Main pages |
6 | 8 |
|
7 | 9 | - [Master Operating Manual](../docs/ai-os/master-operating-manual.md) |
| 10 | +- [Continuous Improvement Loop](../docs/methodology/continuous-improvement-loop.md) |
8 | 11 | - [Mermaid Diagram Catalog](../docs/diagrams/README.md) |
9 | 12 | - [Loop Catalog](../docs/loops/README.md) |
| 13 | +- [Verification Gates](../docs/verifiers/README.md) |
10 | 14 | - [Prompt Templates](../prompts/README.md) |
| 15 | +- [Architecture Guide](Architecture-Guide.md) |
| 16 | +- [Loops](Loops.md) |
| 17 | +- [Verifiers](Verifiers.md) |
| 18 | +- [Prompts](Prompts.md) |
| 19 | +- [Releases](Releases.md) |
11 | 20 |
|
12 | 21 | ## What this is |
13 | 22 |
|
14 | 23 | AI-OS is a repeatable operating framework for AI coding agents. |
15 | 24 |
|
16 | | -The core idea: |
17 | | - |
18 | | -```text |
19 | | -Prompt once. |
20 | | -Verify everything. |
21 | | -Loop until the goal is achieved. |
22 | | -Escalate when risk requires human approval. |
23 | | -``` |
24 | | - |
25 | | -## System overview |
26 | | - |
27 | 25 | ```mermaid |
28 | 26 | flowchart TD |
29 | | - Goal[Goal] --> Agent[AI Coding Agent] |
30 | | - Agent --> Context[Repository Context] |
31 | | - Agent --> Loop[Selected Engineering Loop] |
32 | | - Loop --> Verify[Verification Gates] |
| 27 | + Goal[Goal] --> Context[Load context] |
| 28 | + Context --> Specialists[Fan-out specialists] |
| 29 | + Specialists --> Plan[Plan] |
| 30 | + Plan --> Human{Approval needed?} |
| 31 | + Human -- Yes --> Approval[Human approval] |
| 32 | + Human -- No --> Work[Work] |
| 33 | + Approval --> Work |
| 34 | + Work --> Verify[Verify] |
33 | 35 | Verify --> Done{Done?} |
34 | | - Done -- No --> Loop |
35 | | - Done -- Yes --> Report[Final Report] |
| 36 | + Done -- No --> Work |
| 37 | + Done -- Yes --> Report[Final report] |
36 | 38 | ``` |
0 commit comments