agentleFS
Sign inStart free
naman10parikhnaman10parikh/agentgrid/.claude/skills/troubleshoot/SKILL.md

troubleshoot

Skill9 starsChanged 5 months agoMITView on GitHub

Multi-layer error recovery. Uses docs, web search, context7, learnings, and browser to unblock errors instead of failing repeatedly.

We split this skill into the rules it gives an agent

14 rules. 0 are common, 11 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 17 edited versions we read keep these.

  • 11

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Search memory/LEARNINGS.md for the error message or pattern

    line 14
  2. Distinctive:

    If found: apply the documented fix immediately

    line 15
  3. Distinctive:

    Use context7 MCP to query the relevant library's docs

    line 19
  4. Distinctive:

    Search for the exact error message + library name

    line 24
  5. Distinctive:

    Look for GitHub issues, Stack Overflow, blog posts

    line 25
  6. Distinctive:

    Search awesome-claude-code and similar repos for patterns

    line 30
  7. Very specific:

    Use Grep and Glob to find how others solved the same problem

    line 31
  8. Distinctive:

    Check if there's an MCP server that handles this

    line 32
  9. Distinctive:

    Write a detailed prompt with:

    line 42
  10. Distinctive:

    NEVER retry the exact same approach more than twice

    line 51
  11. Distinctive:

    ALWAYS check learnings first — we may have solved this before

    line 52
  12. Very specific:

    After fixing: add the solution to memory/LEARNINGS.md

    line 53
  13. Distinctive:

    After fixing: check if a skill or hook should be created to prevent recurrence

    line 54
  14. Distinctive:

    When searching: be specific.

    line 56

“Common” is measured: kept by at least 6 in 10 of the 17 edited versions of this file we could read. “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

Troubleshoot — Never Fail the Same Way Twice

Error Recovery Ladder (escalate through levels)

Level 1: Memory Check (10 seconds)

  • Search memory/LEARNINGS.md for the error message or pattern
  • If found: apply the documented fix immediately

Level 2: Docs Check (30 seconds)

  • Use context7 MCP to query the relevant library's docs
  • Example: resolve-library-id("@e2b/desktop") then query-docs for the specific API

Level 3: Web Search (1 minute)

  • Search for the exact error message + library name
  • Look for GitHub issues, Stack Overflow, blog posts
  • Extract the fix and apply it

Level 4: Codebase Search (1 minute)

  • Search awesome-claude-code and similar repos for patterns
  • Use Grep and Glob to find how others solved the same problem
  • Check if there's an MCP server that handles this

Level 5: Alternative Approach (2 minutes)

  • Step back. Is there a completely different way to achieve the same goal?
  • Can you use a different library? Different API? Different pattern?
  • The simplest solution that works is the best solution

Level 6: Claude.ai Prompt (last resort)

  • Write a detailed prompt with:
    • What you're trying to do
    • What you've tried
    • The exact error
    • Relevant code snippets
  • Tell the user: "Paste this into Claude.ai for an answer"

Rules

  1. NEVER retry the exact same approach more than twice
  2. ALWAYS check learnings first — we may have solved this before
  3. After fixing: add the solution to memory/LEARNINGS.md
  4. After fixing: check if a skill or hook should be created to prevent recurrence
  5. Time-box each level. If 2 minutes pass with no progress, escalate.
  6. When searching: be specific. "e2b desktop screenshot timeout" not "e2b error"

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.

Check your setup, free