@haposoft/cafekit 0.7.14 → 0.7.15

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@haposoft/cafekit",
3
- "version": "0.7.14",
3
+ "version": "0.7.15",
4
4
  "description": "Spec-Driven Development workflow for AI coding assistants. Supports Claude Code and Antigravity with spec-first workflows plus Claude Code hapo: skills.",
5
5
  "author": "Haposoft <nghialt@haposoft.com>",
6
6
  "license": "MIT",
@@ -11,13 +11,17 @@ Focus on capabilities and outcomes, not code structure.
11
11
  - Features and capabilities
12
12
  - Domain language and concepts
13
13
  - Data relationships and workflows
14
- **Required (Code-Level Implementation Style)**:
15
- - File paths and specific UI components
16
- - Database tables, columns, and Zod/Type schemas
17
- - API payloads, routes, and JSON contracts
18
- - Edge cases, error handling, and exact validation thresholds
19
-
20
- **Rationale**: AI Coders (like god-developer or ck) consume these tasks directly. If a sub-task is written like PM business jargon (e.g., "Define pass thresholds") instead of code-level instructions (e.g., "Implement `gap_ratio < 15%` threshold in `evaluator.ts`"), the AI coder will hallucinate or fail. Tasks are not PRDs; they are explicit coding orders.
14
+ **Required (Hybrid Human-AI Style)**:
15
+ Every sub-task MUST balance Human Intent (for PM review) and Code-Level Details (for AI implementation).
16
+ Detail bullets must include:
17
+ 1. **Human Intent (The "Why")**: Briefly explain the business logic, expected UX behavior, or why this code exists (e.g., "Mục đích: Chặn user sử dụng extension nếu chưa đồng ý Privacy").
18
+ 2. **AI Code-Level Details (The "How")**:
19
+ - File paths and specific UI components to create/modify.
20
+ - Database tables, columns, and Zod/Type schemas (e.g., `Update users.consent_version`).
21
+ - API payloads, routes, and JSON contracts.
22
+ - Edge cases, error handling, and exact validation thresholds (e.g., `Return 403 if invalid`).
23
+
24
+ **Rationale**: Humans review tasks to verify business requirements are met; AI Coders (like god-developer or ck) read tasks to write explicit code. If you only write business jargon, the AI hallucinates. If you only write code names, the human reviewer cannot verify the business value. You MUST provide both.
21
25
 
22
26
  ### 2. Task Integration & Progression
23
27