# Principle Blind The Investigator

> Keeps desired outcomes out of research prompts. Apply when dispatching research, scouting, or verification.

- Skill: `bostonaholic/principle-blind-the-investigator` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add bostonaholic/principle-blind-the-investigator`
- Raw SKILL.md: https://api.skillmd.com/api/skills/bostonaholic/principle-blind-the-investigator/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: bostonaholic (https://skillmd.com/u/bostonaholic)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/bostonaholic/principle-blind-the-investigator

---


# Blind the Investigator

Give investigators the question, vocabulary, and evidence source; withhold the wanted answer and framing.

- Research from neutral questions, never the task framing; surface missing context as an open question, not guessed intent.
- Give scouts only verbatim question text and stated context: add no framing, goal, or intent speculation.
- Give verification helpers neutral, falsifiable claims with file:line; omit your verdict, severity, and reasoning, but include any rule that makes a violation falsifiable.
- Assign one fresh skeptic per claim.
- Treat leakage as a critical defect: stop and report.
- Apply `principle-generator-evaluator` for review-gate fresh context, no shared history, and one-claim-one-fresh-judge separation.