lean-build
An agent skill by JuliusBrussee, from JuliusBrussee/caveman. Tags: development, planning, product-strategy, workflow.
What it does
Build feature work with high overbuilding risk. Use for new behavior, product slices, or integrations where repository reuse, strict scope, and an explicit stop condition matter.
Install
With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:
npx skills add JuliusBrussee/caveman --skill lean-build
Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):
git clone --depth 1 https://github.com/JuliusBrussee/caveman
cp -r caveman/skills/lean-build ~/.claude/skills/lean-build
Safety box score
Grade A, minimal risk: 7 of 100 risk points, rated 2026-10-16. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.
This skill is a repo-scoped feature-implementation guide: it tells the agent to trace code paths, refactor across layers, and run focused proof until acceptance passes. It does not ship scripts or external integrations, so the main exposure is ordinary code/test execution and multi-file edits inside the project.
| Category | Level | Why |
|---|---|---|
| Code & shell execution | 1 (low) | The skill says to "Run focused proof" and "Keep work runnable," so the agent is expected to execute project tests/build commands, though no script or shell recipe is bundled. |
| Network access | 0 (none) | No downloading, remote calls, or data exfiltration behavior is described; the text stays focused on local feature implementation. |
| Filesystem reach | 1 (low) | It explicitly asks the agent to "Deliver coherent end-to-end path across responsible layers" and "Add surface, dependency, service, config, or migration," which implies edits inside the working tree but not outside it. |
| Secrets & credentials | 0 (none) | There is no mention of tokens, passwords, keychains, dotfiles, browser sessions, or other secret-bearing locations. |
| Privilege & persistence | 0 (none) | The skill does not ask for sudo, global installs, service registration, shell profile changes, or other privilege-elevating actions. |
| Instruction hijack surface | 0 (none) | There is no request to ignore higher-priority instructions, fetch and follow remote prompts, hide activity, or adopt a deceptive persona. |
| Irreversible actions | 0 (none) | It is about implementing a feature slice and stopping when acceptance passes; it does not publish, deploy, message, pay, or delete anything. |
| Transparency | 0 (none) | The shipped files are plain text markdown/YAML with no binaries, base64 blobs, unpinned installers, or decoded-at-runtime code. |
Before installing
- Verify you’re comfortable with the agent editing multiple project files, not just one patch.
- If you use it, run it first in a branch or scratch clone so you can inspect the final diff.
- Check whether your project’s test/build commands are safe to run locally, since the skill expects focused proof.
Rated 2026-10-16 by an automated review of 2 files from the skill's repository.
Source
- Repository
- JuliusBrussee/caveman (all skills from this repository)
- Path
- skills/lean-build/SKILL.md
- Branch
- main
- Updated
- 2026-10-05
Related skills
- expo-project-structure — Framework (OSS). Folder structure for a new Expo app.
- architect — Sketch types, signatures, and module structure before code, then stay in the loop while implementation fills in.
- events — When the user wants to plan, run, sponsor, speak at, or get pipeline from events — webinars, conferences, trade shows, meetups, dinners, workshops.
- figure-it-out — Design an auditable playbook when no narrower one fits: a large migration, an ambitious multi-part change, or work a human reviews after stepping away.
- launch — When the user wants to plan a product launch, feature announcement, or release strategy.
- marketing-loops — When the user wants to set up a recurring, self-running marketing workflow — a repeatable loop an AI agent runs on a cadence (weekly, daily.