Skip to content
Get packs

Research Implement Feature

Builds a working prototype from a plain request, logging every open decision as it goes

Source Download skill
Repository
wanshuiyin/Auto-claude-code-research-in-sleep
Stars
18,482
Licence
MIT
Files
1 file

What it does

When you ask an agent to just build something (for example "implement X for me"), this skill makes it first produce a minimal version that runs end to end, then add one feature at a time, each with its own check. Every choice the request left open is written into an assumption ledger before the code depends on it, and a second model family is then used to look for assumptions that were not declared.

Great for

  • Developers who want a quick working prototype from a capability description without being interviewed first
  • Researchers turning a decided idea into runnable code for an experiment
  • Teams that want an auditable list of the assumptions an agent made during a build

Skip it if

  • Open-ended research direction finding, where the research-pipeline skill is the better fit (the skill decides nothing of consequence silently, so it expects a decided target)
  • Work where unreviewed autonomous decisions are risky, since the default mode proceeds without asking and the author must review the ledger and diff afterwards
  • Runs that need the reviewer tools (Codex, Antigravity, Grok, Claude review) if those are not configured, since reviewer slots can be unavailable

Try asking

  • “Implement a CLI tool that converts my CSV files to JSON, prototype first then add validation”
  • “Build this feature: a small web page that shows live metrics, get a running version first”
  • “Prototype then extend a text-summarisation helper and log every assumption you make”
  • “Give me a working implementation of this capability description, using the base repo at the URL I provide”
Step-by-step guide: how to use Research Implement FeatureInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    research-implement-feature.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip research-implement-feature.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip research-implement-feature.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • State the capability clearly and mention any base repo with base repo: <url> so the agent matches existing conventions
  • Use ask: semantic if you want to be asked about decisions that change the meaning of results, rather than reviewing them afterwards
  • Read the assumption ledger and the diff after the run, since the agent may have installed dependencies or chosen defaults on your behalf

What's inside

  • SKILL.md

Redistributed unmodified from wanshuiyin/Auto-claude-code-research-in-sleep under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.