research

Skill: Research

Use When

Use this skill for deep research tasks that require:

• structured problem framing
• state-of-the-art mapping
• historical evolution and future direction mapping
• falsifiable hypothesis generation
• explicit uncertainty and evidence attribution
• adversarial challenge and replication planning

Inputs

• Required:
  • research question, decision, or hypothesis to investigate
  • enough context to define scope, audience, and constraints
• Optional:
  • seed sources, prior notes, or local research_vault/ material
  • target output format, deadline, or calibration horizon

Workflow

1. Follow the full protocol in references/research_protocol.md.
2. Treat references/system_prompt.md as behavioral constraints.
3. Use local memory at ./research_vault/ when available.
4. If filesystem write is restricted, return Simulated Vault Output and state limitations.

Output Contract

Use exact section titles:

1. Core Essence
2. Evolutionary Map
3. Vulnerabilities and Blind Spots
4. Cross-Boundary Inspiration
5. Falsifiable Conclusions

Append:

• Source Table
• Uncertainties
• Self-Audit
• Priority Matrix
• Predictions and Calibration Plan
• Adversarial Review
• Replication Pack
• Vault Writeback Log

Install

From repository root:

# Codex (default)
./scripts/install_skill.sh research

# Claude Code
./scripts/install_skill.sh research --tool claude

# Other tools
./scripts/install_skill.sh research --dest ~/.my-tool/skills

Limits

• This skill is designed for rigorous synthesis, not instant answers.
• Output quality depends on source quality, scope clarity, and available evidence.
• Common failure cases:
  • weak or missing primary sources
  • underspecified scope or success criteria
  • missing filesystem access for real vault writeback