🔍
AI Engineering

Research

by JuliusBrussee

Research is an AI Engineering skill for Claude Code, published by JuliusBrussee in cavekit.

1.1K stars86 forkson JuliusBrussee/cavekitAdded 2026/08/20Repository updated 2026/08/14
claude-codeparallel-agentsskillsspec-driven-developmenttest-driven-development
Install in seconds
Install Research
Copy Research into your Claude Code skills folder. Run the command in your terminal, or review the source on GitHub before installing.
terminal
npx degit https://github.com/JuliusBrussee/cavekit/tree/main/skills/research ~/.claude/skills/research

Requires Node.js. Downloads this skill only — not the rest of the repository — into your Claude Code skills folder.

Without Node.js

git clone https://github.com/JuliusBrussee/cavekit.git

Clones the whole repository, then copy the skill’s own directory into your skills folder yourself.

In this catalog

Source file
skills/research/SKILL.md in JuliusBrussee/cavekit
Installs to
~/.claude/skills/research
Collection
One of 9 skills cataloged from this repository
Category
AI Engineering2451 skills

What Research does

Research gathers external knowledge and logs it as source-cited §R rows so builds ground in facts instead of hallucinating library behavior. Use it when a spec decision hinges on a library, API, or best practice you are unsure of.

Research is cataloged under AI Engineering on DirSkills. Research comes from a repository tagged claude-code, parallel-agents, skills, spec-driven-development and test-driven-development.

Documentation

README

research — external knowledge → §R

Every finding cites a source. No source → flag it ?, never write a guess as fact.

"Process without library context gives you well-organized hallucinations." Build invents a plausible-but-wrong API & §B fills with avoidable bugs. Research is the external oracle: pull the real fact once, log it caveman, never re-derive.

WHEN TO RESEARCH

  • A §C/§I/§V decision hinges on a lib, API, version, or pattern you are unsure of.
  • You are about to assume how an external dependency behaves.
  • The idea touches a domain with real prior art (auth, payments, crypto, rate-limit).
  • /grill parked a ? that the outside world must answer.

Skip when the build touches only code you already wrote. Research scales to the unknown, ⊥ to habit.

This is the opening of the README. Read the full README on GitHub.

Commands Research provides

Slash commands named in this skill’s SKILL.md, listed in the order they first appear.

  • /grill

Frequently asked about Research

  • What else does JuliusBrussee publish alongside Research?

    Research is one of 9 skills that DirSkills catalogs from JuliusBrussee/cavekit, the repository it ships in. Its siblings there include Adversarial Spec Review, Backprop and Build. Each one is a separate skill with its own page in this directory, installs the same way Research does, and is maintained by JuliusBrussee in that same repository. The rest of the collection is listed on the JuliusBrussee/cavekit page.

  • How does Research compare to other AI Engineering skills?

    Research ranks #1237 by stars among the 2451 AI Engineering skills in this catalog. The most-starred ones next to it are Architecture Decision Records, AI-First Engineering and Agentic OS. DirSkills ranks by the star count of the repository each skill ships in, so that order reflects how popular those repositories are rather than any review of Research against them. Open each page to compare what they document and how they install.

More from JuliusBrussee/cavekit

Research is one of 9 skills cataloged on DirSkills from JuliusBrussee/cavekit.

See all 9 skills
🧐
2w ago

Adversarial Spec Review

Adversarial Spec Review constructs a skeptical reviewer that attacks a spec before any code is written, using evidence from the codebase, research, and live best-practices. Use it before high-blast-radius builds to refute unsound plans and harden verification.
Quality
1.1K86
🐞
2w ago

Backprop

Backprop traces a bug or failing test to its root cause, then appends a §B entry and usually a §V invariant to SPEC.md so recurrence is impossible. Use it after test failures, bug reports, post-mortems, or VIOLATE flags.
Quality
1.1K86
🏗️
2w ago

Build

Build implements tasks from a SPEC.md using a single-threaded plan-then-execute loop. It tracks task status, runs the verification command, and on failure invokes backprop to update spec invariants before retrying.
Quality
1.1K86
🗿
2w ago

Caveman

Caveman encodes SPEC.md and spec-adjacent writes into compressed symbolic notation, cutting tokens ~75% versus prose while preserving facts. Use it when writing or editing specs, backprop entries, or whenever brevity is requested.
Writing
1.1K86
🧹
2w ago

Deepen

Deepen finds the shallowest modules in the code the spec touches, researches a deeper design, and proposes refactors that shrink interfaces and hide decisions while keeping behavior constant and tests green before and after.
Quality
1.1K86
🧐
2w ago

Grill

Grill calibrates interrogation of a fuzzy idea before it becomes a spec, asking one question at a time and landing answers as goals or constraints. Use it when a plan is vague or needs stress-testing before writing a spec.
Quality
1.1K86