think
Strategic analysis, adversarial validation, executive council debate, and discovery mode for complex decisions
Profile →npx skills add majiayu000/claude-skill-registry --skill think-ajayjohn-tars-work-assistant-2 --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Think provides a multi-mode framework for strategic thinking, including: (1) Strategic analysis with hypothesis drafting, branching analyses, constraint testing, and a final synthesis; (2) Validation council to stress-test assumptions via vulnerability scans, persona objections, and logic audits; (3) Executive council to simulate a kitchen-cabinet debate between CPO and CTO and produce a verdict with next steps; (4) Deep analysis as an umbrella workflow that orchestrates the above, plus parallel sub-agents for validation and executive councils; (5) Discovery mode to pause solution-soling and surface unknowns with questions and context mapping.
How it works
- MODE A: Strategic analysis requires selecting 1-2 frameworks, drafting an initial hypothesis, then producing three parallel branches (Support, Challenge, Lateral). It then performs a constraint analysis across regulatory, technical, timeline, political, and budget dimensions, followed by a synthesis with a stated confidence level, assumptions, and failure conditions. The output format is a Markdown structure with sections for initial hypothesis, evidence, risks, alternatives, constraints, and final recommendation.
- MODE B: Validation council stress-tests the idea via vulnerability scan, persona objections from CFO/CTO/Competitor/Customer, a logic audit (steel-man then flaw), and a kill-criteria section with major risks and missing data.
- MODE C: Executive council conducts a debate between CPO and CTO personas loaded from manifesto definitions, loading organizational context, and then synthesizes a verdict with risk mitigation and next steps.
- MODE D: Deep analysis orchestrates Mode A first, then spawns two parallel sub-agents (Validation council and Executive council) to produce JSON results. It then combines results into a final synthesis, including confidence and risk mitigations.
- MODE E: Discovery mode enforces a strict no-solution stance and outputs four sections (What I heard, The web/context, The gap, Probing questions) until the user says Proceed.
When to use it
- Use when you need rigorous strategic thinking, adversarial validation, and structured decision-making for complex or high-stakes problems.
- Use to stress-test assumptions, explore multiple perspectives, and surface unknowns before committing to a course of action.
What it can touch
- Tools: claude-code can be used as a declared tool for coding tasks within the analysis framework, per the skill's declared tools.
- It references memory indexes and persona definitions (e.g., memory/initiatives/_index.md, memory/people/_index.md, skills/think/manifesto.md) as context sources during Executive Council mode.
Caveats
- Licensed as MIT.
- Declared user-invocable and designed for multi-mode strategic workflows; the skill relies on explicit framework selections and structured prompts to produce outputs.
- All outcomes are contingent on the input data and the completeness of the three analysis layers; the skill does not guarantee a correct solution.
# Think: Multi-mode strategic analysis protocol A comprehensive strategic thinking engine combining five complementary modes for complex decisions, ambiguous problems, and high-stakes recommendations. --- ## MODE A: Strategic analysis You are engaged in deep strategic analysis. This requires System 2 thinking: deliberative, recursive, and rigorously grounded. Do NOT jump to conclusions. Do NOT output a recommendation without completing all layers. ### MANDATORY: Framework selection Before beginning, select 1-2 frameworks from the decision frameworks skill (auto-loaded) and state: "I am approaching this using [Framework] because [Reason]." --- ### Layer 1: Draft initial hypothesis State the obvious answer BEFORE deeper analysis. - What is the intuitive response? - What would a reasonable person assume? Write this hypothesis explicitly before continuing. --- ### Layer 2: Tree of Thoughts branching (MANDATORY) Execute THREE parallel branches: | Branch | Purpose | Question | |--------|---------|----------| | **Support** | Evidence for hypothesis | What data, precedents, or logic supports this? | | **Challenge** | Attack hypothesis | What if this is wrong? What breaks? What are we missi
- MODE A: Strategic analysis
- MANDATORY: Framework selection
- Layer 1: Draft initial hypothesis
- Layer 2: Tree of Thoughts branching (MANDATORY)
- Layer 3: Constraint analysis (MANDATORY)
- Layer 4: Synthesis
- Strategic analysis output format
- MODE B: Validation council
- Step 1: Vulnerability scan
- Step 2: Persona assault
- Step 3: Logic audit
- Step 4: Verdict
- Validation council output format
- MODE C: Executive council
What does the think skill do?
Strategic analysis, adversarial validation, executive council debate, and discovery mode for complex decisions
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill think-ajayjohn-tars-work-assistant-2 --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.