agentleFS
claude-worldclaude-world/director-mode-lite/skills/focus-problem/SKILL.md

focus-problem

Skill80 starsChanged 8 months agoMITView on GitHub

Deeply scope a problem before coding: parallel Explore agents map related files, similar implementations, and test patterns, producing a problem definition, boundaries, and minimal implementation plan. Use when planning a feature, investigating a bug, or entering an unfamiliar codebase area.

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before diving in, use the Explore agent to understand the context:

    line 13
  2. Very specific:

    Before dispatch, preserve the bounded exploration question and expected result.

    line 15
  3. Very specific:

    If Claude's Agent tool is unavailable or withheld, or Codex's spawn_agent interface is unavailable or withheld, the maximum depth is …

    line 16
  4. Distinctive:

    Follow this process strictly.

    line 92

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Focus Problem Analysis

Analyze: $ARGUMENTS

Explore Codebase First

Before diving in, use the Explore agent to understand the context:

Before dispatch, preserve the bounded exploration question and expected result. If Claude's Agent tool is unavailable or withheld, or Codex's spawn_agent interface is unavailable or withheld, the maximum depth is reached, or a concurrency limit rejects the call, do not retry or invent exploration findings. Return an incomplete dispatch_request and do not advance the plan.

Choose exactly one provider-native form:

  • Claude Code: Agent(subagent_type="Explore", model="haiku", prompt="<bounded prompt>")
  • Codex CLI: spawn_agent(agent_type="explorer", task_name="focus_problem", message="<bounded prompt>")
Agent(subagent_type="Explore", model="haiku", prompt="""
Explore the codebase for: $ARGUMENTS (thoroughness: medium)
Focus on:
1. Relevant files and modules
2. Existing similar features
3. Test patterns
4. Architecture patterns
""")

In Codex, send the same exploration request through spawn_agent(...); do not invoke the Claude example as a second dispatch.

Analysis Framework

1. Core Problem Identification

  • What is the real problem?
  • Why does it need to be solved?
  • What happens if we don't solve it?

2. Minimal Solution Scope

  • What's the minimum change needed?
  • Which features are essential?
  • What can be deferred or omitted?

3. Solution Evaluation

  • Option A (Simplest):
  • Option B (Balanced):
  • Option C (Complete):
  • Recommendation: Choose the simplest viable option

4. Implementation Plan

  • Step 1: Write tests
  • Step 2: Minimal implementation
  • Step 3: Update documentation
  • Step 4: Commit changes

Output Template

## Problem Definition

### Summary
[One sentence description]

### User Need
[Who needs this? What do they want to achieve?]

### Success Criteria
[How do we know it's done?]

### Scope
- In scope: [List]
- Out of scope: [List]

### Affected Files
- [file1.ts] - [why]
- [file2.ts] - [why]

### Risk Assessment
- [Potential risks and mitigations]

Follow this process strictly. Don't skip steps.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.