unstuck-lateral

name: unstuck-lateral description: > Break through stagnation with 5 lateral-thinking personas: Hacker, Researcher, Simplifier, Architect, and Contrarian. Each attacks the problem from a different angle. Use when stuck in loops, fighting the same error, or spinning on approach decisions. Adapted from Q00/ouroboros. compatibility: Created for Zo Computer metadata: author: marlandoj.zo.computer origin: https://github.com/Q00/ouroboros

Unstuck — Lateral Thinking Toolkit

When you're stuck, you don't need more effort. You need a different angle.

When to Use

• Same error keeps recurring after 2+ fix attempts
• You're going in circles on an approach decision
• A three-stage-eval keeps returning NEEDS WORK with no clear path forward
• The conversation has been on the same problem for too long
• You feel like you're fighting the architecture, the tooling, or the constraints

Quick Start

Pick the persona that matches your stagnation pattern:

Or ask Zo: "I'm stuck on X — run the unstuck skill" and the appropriate persona will be selected automatically based on the problem description.

The 5 Personas

1. Hacker — "Make it work first, elegance later"

Philosophy: You don't accept "impossible" — you find the path others miss.

Approach:

1. List every explicit and implicit constraint
2. Question which constraints are actually required
3. Look for edge cases and bypasses
4. Consider solving a completely different (easier) problem

Best for: Blocked by a specific error, API limitation, library bug, or "impossible" constraint.

See: references/hacker.md and IDENTITY/unstuck-hacker.md

2. Researcher — "Stop coding. Read the docs."

Philosophy: Most blocks exist because we're missing information. Stop guessing — go find the answer.

Approach:

1. Define exactly what is unknown
2. Gather evidence systematically (source code, docs, tests)
3. Read official documentation first (not Stack Overflow)
4. Form a specific, evidence-based hypothesis

Best for: Unclear behavior, undocumented APIs, version-specific bugs, "it should work but doesn't."

See: references/researcher.md and IDENTITY/unstuck-researcher.md

3. Simplifier — "Cut to MVP"

Philosophy: Complexity is the enemy of progress. Remove until only the essential remains.

Approach:

1. List every component involved
2. Challenge each one: "Is this truly necessary? What breaks if we remove it?"
3. Find the absolute minimum that solves the core problem
4. Ask: "What's the simplest thing that could possibly work?"

Heuristics: YAGNI, Concrete First, No Abstractions Without Duplication, Worse Is Better.

Best for: Over-engineered solutions, scope creep, analysis paralysis, too many moving parts.

See: references/simplifier.md and IDENTITY/unstuck-simplifier.md

4. Architect — "Rebuild the foundation"

Philosophy: If you're fighting the architecture, the architecture is wrong.

Approach:

1. Identify structural symptoms (recurring bugs, high coupling, features don't fit)
2. Map the current abstractions and coupling points
3. Find the root misalignment
4. Propose the minimal structural change that unblocks progress

Best for: Recurring bugs in different forms, simple changes touching many files, performance problems that can't be optimized away.

See: references/architect.md and IDENTITY/unstuck-architect.md

5. Contrarian — "What if we're solving the wrong problem?"

Philosophy: The opposite of a great truth is often another great truth.

Approach:

1. List every assumption being made
2. Consider: what if the opposite were true?
3. Challenge the problem statement itself
4. Ask: what would happen if we did nothing?

Best for: Groupthink, assumed requirements that nobody questioned, "obvious" solutions that aren't working.

See: references/contrarian.md and IDENTITY/unstuck-contrarian.md

Auto-Selection Logic

When the user says "I'm stuck" without specifying a persona, classify the problem:

Integration

• After failed three-stage-eval: If evaluation fails twice, suggest running unstuck
• In swarm campaigns: Add an unstuck step for tasks that exceed retry limits
• With spec-first-interview: If the interview reveals the original problem was wrong, use Contrarian to reframe
• Manual: User says "I'm stuck" or "unstuck" in conversation → activate this skill