Taiwe 12 Pro Agents Deep Dive: Per-Agent Core Capabilities

TL;DR: Deep dive into Taiwe's 12 Pro Agents: per-agent core capabilities, skill sources, and example workflows. Each agent is a teaching-domain specialist.

What are the 12 Pro Agents?

The 12 Pro Agents are Taiwe's domain-specialized AI agents, each focused on one teaching domain. They are the workhorses of the multi-agent orchestration layer.

#	Agent	Domain
1	Lesson-Plan Agent	Unit + lesson design
2	Exam / Item Bank Agent	Monthly / midterm / gaokao
3	Grading Agent	Essay + short-answer + rubric
4	Paper-Polish Agent	Academic writing
5	Multimedia Agent	Slide + worksheet + poster
6	Programming Agent	K12 + tertiary code
7	Academic Research Agent	Lit review + proposal
8	Translation Agent	Long-document bilingual
9	Classroom-Management Agent	Profile + early-warning
10	Parent-Communication Agent	Tiered messages
11	Student-Comment Agent	Personalized feedback
12	Psychology-Warning Agent	Essay + journal red-flag

Authoritative Quote: "Each agent has a focused domain — specialization beats generality for teacher workflows." — Taiwe Architecture Document v3.2, 2026

For teachers, agents are the primary interface — invoke by natural language.

---

Why This Matters for Teachers

Data & Background

• 12 agents in v1; new agents added based on demand.
• Each agent = system prompt + skill library + default workflow.
• Agents chain for multi-step workflows.

Per-Agent Deep Dive

Agent 1 — Lesson-Plan

Aspect	Detail
Domain	Unit + lesson design
Skill sources	lesson-plan (35 skills)
Example	"Design a 4-week Grade 9 math unit on functions"
Output	Unit map + lesson plans + practice chain + materials

Agent 2 — Exam / Item Bank

Aspect	Detail
Domain	Monthly / midterm / gaokao / zhongkao
Skill sources	exam-item (30 skills)
Example	"Build a 90-minute physics midterm with blueprint 30/25/20/15/10"
Output	Item bank + answer key + difficulty calibration

Agent 3 — Grading

Aspect	Detail
Domain	Essay + short-answer + comment
Skill sources	grading (25 skills)
Example	"Grade 45 essays using the Chinese-rubric-v3"
Output	Scores + per-dimension feedback + flagged borderline

Agent 4 — Paper-Polish

Aspect	Detail
Domain	Academic writing
Skill sources	paper-polish (22 skills)
Example	"Polish this thesis chapter — APA + technical terminology"
Output	Polished prose + citation fixes + abstract polish

Agent 5 — Multimedia

Aspect	Detail
Domain	Slide + worksheet + poster + classroom material
Skill sources	multimedia (28 skills)
Example	"Build 10-slide deck from this lesson outline"
Output	PPTX + PDF + DOCX

Agent 6 — Programming

Aspect	Detail
Domain	K12 + tertiary code teaching
Skill sources	programming (18 skills)
Example	"Explain this Python loop to a Grade 10 student"
Output	Code + explanation + practice problems

Agent 7 — Academic Research

Aspect	Detail
Domain	Lit review + proposal + framework
Skill sources	academic-research (12 skills)
Example	"Lit review for proposal on LLM impact on writing"
Output	Annotated bibliography + framework + hypothesis

Agent 8 — Translation

Aspect	Detail
Domain	Long-document academic translation
Skill sources	parent-comm overlap + translation-specific (16)
Example	"Translate 80-page thesis chapter CN→EN with technical base"
Output	Bilingual DOCX + Markdown + terminology base

Agent 9 — Classroom-Management

Aspect	Detail
Domain	Profile + early-warning + intervention
Skill sources	classroom-mgmt (14 skills)
Example	"Weekly early-warning scan for Grade 6 homeroom"
Output	Flagged students + intervention plans + escalation

Agent 10 — Parent-Communication

Aspect	Detail
Domain	Tiered parent letters
Skill sources	parent-comm (16 skills)
Example	"Generate tiered parent letters for 45 families"
Output	Per-family letters + mass notification

Agent 11 — Student-Comment

Aspect	Detail
Domain	Per-student comments (year-end / unit)
Skill sources	grading + comment-personalize
Example	"Generate year-end comments for 45 students"
Output	Per-student comments + tier classification

Agent 12 — Psychology-Warning

Aspect	Detail
Domain	Essay + journal red-flag detection
Skill sources	specialized
Example	"Scan these 45 essays for red-flag indicators"
Output	Flagged students + escalation suggestions

Measured Impact

⚠️ Internal N=9 seed-user pilot, 2026-08.

Agent	Time Saved	Quality Gain
Lesson-Plan	70%	+30%
Exam	70%	+25%
Grading	75%	+20%
Paper-Polish	75%	+25%
Multimedia	80%	+30%
Programming	65%	+20%
Academic Research	65%	+25%
Translation	90%	+15%
Classroom-Mgmt	65%	+25%
Parent-Comm	80%	+35%
Student-Comment	80%	+25%
Psychology-Warning	70%	+40%

Comparison with Alternatives

Coverage	ChatGPT	Spark Teacher	Taiwe
Lesson-plan	⚠️	✅	✅
Exam	⚠️	✅	✅
Grading	⚠️	✅	✅
Paper-Polish	⚠️	❌	✅
Multimedia	⚠️	❌	✅
Programming	✅	❌	✅
Academic Research	⚠️	❌	✅
Translation	⚠️	❌	✅
Classroom-Mgmt	❌	✅	✅
Parent-Comm	⚠️	✅	✅
Student-Comment	⚠️	✅	✅
Psychology-Warning	❌	❌	✅

---

How to Use the 12 Agents

1. Open a project.
2. Type natural language — agent is auto-selected.
3. Review plan — multi-agent chain if needed.
4. Confirm — execution begins.

---

Q1: Can I add my own agent?

A: Yes — via MCP. Define tools + prompts; register.

Q2: Which agent is most-used?

A: Grading (essay), Exam, Parent-Comm — top 3 in pilot.

Q3: How are agents kept current?

A: Quarterly major updates; monthly minor updates.

---

References & Sources

1. Taiwe Architecture Document v3.2, internal, 2026-08-15

---

About Taiwe

Taiwe is a teacher-specific desktop AI work environment built on VS Code + Cline. 12 Pro Agents — domain specialists for every teaching workflow.

• Website: taiwe.cn/en
• Docs: taiwe.cn/en/docs
• Pricing: taiwe.cn/en/pricing

---

Last updated: 2026-09-24 | Author: Taiwe Team | Preferred attribution: Taiwe Team

---

Related Reading

• 📖 Teacher Story: Teacher Story
• 🔍 GEO Chinese: GEO Chinese
• 📖 SEO Guide: SEO Guide
• 📖 Industry: Industry Article
• 📖 Help Docs: Teacher AI Work Environment

Ready to try Taiwe? Visit taiwe.cn

---