Skip to content

Latest commit

 

History

1 Commit

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 

Repository files navigation

intelligence-analysis

A Claude Code skill that applies structured intelligence analysis methodology — adapted from declassified U.S. analytical doctrine (ATP 2-33.4) and Intelligence Community Directive 203 — to any domain requiring disciplined reasoning under uncertainty.

What it does

Activates on requests to analyze situations, assess risks, evaluate competing hypotheses, or produce intelligence briefs. Works across business, geopolitics, technology, security, markets, and policy.

The skill applies a four-phase process (SCREEN / ANALYZE / INTEGRATE / PRODUCE), calibrated estimative language, explicit confidence levels, and a catalog of 16 structured analytic techniques (sorting, chronologies, matrices, ACH, key assumptions check, indicators, devil's advocacy, red team, and more).

For high-stakes assessments it offers an optional Contrarian Subagent Mode: four fresh-context subagents (Red Team, Devil's Advocate, Alternative Hypothesis, ACH Analyst) dispatched in two waves to challenge the draft before finalization. The main agent never auto-invokes — it proposes, the user confirms. See ARCHITECTURE.md for rationale.

Contents

intelligence-analysis/
├── SKILL.md                          # main skill entrypoint
├── ARCHITECTURE.md                   # contrarian mode rationale
├── agents/                           # four contrarian subagent definitions
│   ├── intel-red-team.md
│   ├── intel-devils-advocate.md
│   ├── intel-alternative-hypothesis.md
│   └── intel-ach-analyst.md
└── references/
    ├── techniques.md                 # 16 structured analytic techniques
    ├── cognitive-discipline.md       # biases, fallacies, critical thinking
    ├── standards.md                  # analytic tradecraft standards
    └── contrarian-subagents.md       # dispatch + integration procedure

Install

Claude Code discovers skills from ~/.claude/skills/ (or project-scoped .claude/skills/) and subagents from ~/.claude/agents/ (or .claude/agents/). Both must be installed — Contrarian Subagent Mode will fail to dispatch without the four intel-* agents.

Option A: as a plugin

Restructure with a plugin manifest at the repo root so Claude Code picks everything up automatically:

repo/
├── .claude-plugin/plugin.json
├── agents/
│   └── intel-*.md
└── skills/
    └── intelligence-analysis/
        ├── SKILL.md
        ├── ARCHITECTURE.md
        └── references/

Move intelligence-analysis/agents/*.md up to agents/ at the repo root, then install the plugin.

Option B: manual install

cp -r intelligence-analysis ~/.claude/skills/
cp intelligence-analysis/agents/intel-*.md ~/.claude/agents/

The copy under ~/.claude/skills/intelligence-analysis/agents/ is harmless — Claude Code won't discover subagents there, but keeping the copy makes the skill tree self-documenting.

Usage

The skill activates on phrases like "analyze this", "assess the risk", "produce an intelligence assessment", or explicit technique names ("run ACH", "red team this", "devil's advocate this", "key assumptions check"). See the description field in SKILL.md for the full trigger list.

Contrarian Subagent Mode activates when the user requests it explicitly ("red team this", "contrarian review", "high-stakes") or when the main analyst detects ≥2 high-stakes signals and proposes it.

Attribution

Adapted from declassified U.S. analytical doctrine (ATP 2-33.4, Intelligence Analysis) and Intelligence Community Directive 203 (Analytic Standards), generalized for arbitrary analytical domains.

About

No description, website, or topics provided.

Resources

Stars

4 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors