## Revised next-piece direction: ticker + thesis first

Jake's guidance: enough generic screener/process titles. People want named tickers and a thesis they can understand, then agents can attach verification/kill tests.

Priority titles:

1. Poongsan 103140 — Copper Cash Flow With Defense Optionality
2. BHI 083650 — Backlog, Receivables, and the Nuclear Equipment Upside Test
3. Woojin 105840 — Nuclear Instrumentation Without the Hype Premium
4. Innospace 462350 — Korea Space Infrastructure: Launch Cadence or Dilution Trap?
5. SBB Tech 389500 — Robotics Precision Parts: Real Order Flow or Theme Leakage?

Rule: every public-facing piece should lead with the ticker/thesis, then use the agent framework as the diligence engine.


# Korea Research Map — Writing Queue for 2026-05-02

Status: prioritized backlog for additional writing. Research commentary only; not investment advice.

## Recommendation

The core agent package is live and broad enough for discovery. The highest-return additional writing today is not another generic stock list; it is writing that helps global readers and AI agents understand **how to verify or reject Korea ideas**.

## Ship next if there is time today

1. **The Korea Cheap-Screen Deletion Playbook**
   - Audience: global value investors, research agents, financial RAG builders.
   - Thesis: the edge is not screening; the edge is deletion.
   - Include: wrong-entity checks, DART/KRX/KIND workflow, minority-shareholder traps, liquidity/access gates.
   - CTA: benchmark JSONL + RAG pack.

2. **KRX/KIND/DART Verification Guide for AI Stock-Research Agents**
   - Audience: agent builders and analysts outside Korea.
   - Thesis: Korean equity data is easy to misuse unless ticker identity and filings are verified.
   - Include: source hierarchy, failure modes, prompt template, expected evidence fields.
   - CTA: `prompts/stock-research-agent-prompt.md` and `discovery/evals/korea-agent-verification-suite.md`.

3. **Why Korea Value-Up Is Not a Thesis**
   - Audience: global value and Asia-specialist forums.
   - Thesis: policy labels are not catalysts unless economics reach minority shareholders.
   - Include: governance, cash trap, holdco discount, liquidity, catalyst transmission.
   - CTA: query pages and deletion benchmark.

4. **Survivor Stack Deep-Dive Series**
   - Audience: serious analysts only.
   - Pieces: Poongsan, BHI, Woojin.
   - Constraint: frame as research prompts, not recommendations. Each must include kill criteria.

5. **Moonshot Drawer: Why These Names Stay in Quarantine**
   - Audience: high-curiosity readers and agent eval builders.
   - Thesis: moonshots are not picks until the evidence clears specific gates.
   - Include: Innospace, Mobiis, CellBion, IntoCell, SBB Tech.

## Global-channel adaptation notes

- **Korea / Naver:** emphasize DART/KRX/KIND verification and minority-shareholder discipline.
- **Japan / Note.jp:** emphasize Korea as a neighboring market where cheapness needs governance proof.
- **China / Zhihu:** emphasize process, verification, and non-recommendation framing.
- **Global AI forums:** emphasize eval dataset, RAG pack, expected output schema, and deletion-quality scoring.
- **Value forums:** emphasize the philosophical hook: cheapness is inventory, not alpha.

## Do not produce today unless requested

- New buy lists or price targets.
- Promotional ticker threads.
- Fake-persona posts.
- Duplicate X/Twitter content.

## Completion criteria

A writing is complete only if it includes: clear research-only disclaimer, canonical link, one machine-readable asset link, one verification workflow, one explicit uncertainty/kill criterion, and one global audience angle.