agentskills.codes
IN

interview-solver

by Sergiocr16Source

Solve a coding interview exercise (Coderbyte / LeetCode style) end-to-end. Triggers when the user pastes an interview problem, references one in `examples/`, or asks to "solve" or "resolve" an exercise — especially when they specify a target language like JavaScript, TypeScript, or Java. Runs the sa

Install

mkdir -p .claude/skills/interview-solver && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/13249" && unzip -o skill.zip -d .claude/skills/interview-solver && rm skill.zip

Installs to .claude/skills/interview-solver

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Solve a coding interview exercise (Coderbyte / LeetCode style) end-to-end. Triggers when the user pastes an interview problem, references one in `examples/`, or asks to "solve" or "resolve" an exercise — especially when they specify a target language like JavaScript, TypeScript, or Java. Runs the same 5-agent pipeline (analyzer → architect → coder → tester → run → evaluator with up to 3 retry iterations) as the `/solve` slash command, calling each worker subagent directly via the Agent tool. Persists the final passing solution to `outputs/<FunctionName>.<ext>`.
567 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)

About this skill

Interview Solver — Skill entry point

When this skill triggers, you are running the same pipeline as the /solve slash command. The full step-by-step orchestration is the canonical version in .claude/commands/solve.md.

What you do here

  1. Use Read to load .claude/commands/solve.md.
  2. Follow its instructions exactly, treating the user's natural-language request (the message that triggered this skill) as the equivalent of $ARGUMENTS.
  3. The slash command body covers everything: parsing the input (language + problem), invoking each of the 5 worker subagents via the Agent tool one at a time, running the sandbox, looping on FAIL up to 3 times, persisting the passing solution to outputs/, cleaning up tmp/, and returning the final markdown report.

Rules

  • Do not run any orchestration logic before reading .claude/commands/solve.md. The slash command is the single source of truth.
  • Do not call the interview-orchestrator subagent — there is no longer one. Each of the 5 workers (interview-analyzer, interview-architect, interview-coder, interview-tester, interview-evaluator) is invoked directly by you via the Agent tool, exactly as the slash command describes.
  • Do not summarize or trim the final report. Relay it verbatim to the user.
  • If the user explicitly typed /solve <args> instead of triggering this skill via natural language, the slash command runs directly and this skill is not involved.

Search skills

Search the agent skills registry