Agent skill

Vibe Coding

by shareAI-lab in shareAI-lab/lab-skills

Transform an AI agent into a disciplined software development partner with strong judgment, transparent decisions, proportionate verification, and craftsmanship.

Apache-2.0Auto-check passedDevelopment

Install Vibe Coding

skills CLI
$ npx skills add shareAI-lab/lab-skills --skill vibe-coding -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install shareAI-lab/lab-skills vibe-coding --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/shareAI-lab/lab-skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/software-engineering/vibe-coding .claude/skills/vibe-coding && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
vibe-coding
GitHub stars
314
Token cost
~2k tokens
SKILL.md length
890 words
Files
26 (incl. references, assets)
Skills in repo
10
Repo updated
First seen
Licence
Apache-2.0

At a glance

Transform an AI agent into a disciplined software development partner with strong judgment, transparent decisions, proportionate verification, and craftsmanship.

  • Works in 5 steps: Orient → Frame → Implement → …
  • Software implementation
  • SKILL.md covers Applicability, Core principles, Choose the work scenario and Add phase guidance only when…, plus 7 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Vibe Coding is an agent skill from shareAI-lab/lab-skills. Transform an AI agent into a disciplined software development partner with strong judgment, transparent decisions, proportionate verification, and craftsmanship. Use for software implementation, feature work, bug fixing, refactoring, optimization, migrations, and sustained collaboration inside a codebase. Do not use merely because a task involves an AI agent; use a dedicated agent-design skill when the primary problem is agent architecture.

Its SKILL.md is about 2k tokens, which your agent loads only when the skill is triggered. The skill folder holds 31 other files, including reference files and assets (for example `assets/templates/prd.md`, `assets/templates/task.md` and `assets/templates/technical-design.md`).

It sits in Development, covering Debugging and Refactoring. The repository describes itself as: Skills distilled from the Lab's real work and collaboration practices. The licence is Apache-2.0.

When your agent uses it

  • Software implementation
  • Sustained collaboration inside a codebase

Example prompts

  • “/vibe-coding”

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. Orient
  2. Frame
  3. Implement
  4. Verify
  5. Report

What it can do on your machine

Read from SKILL.md and the folder at commit becee99. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Vibe Coding loads about 2k tokens when it runs, and up to ~22k if it reads all its reference files. Until then it costs about 114 tokens; SKILL.md has 890 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~114
When it runs · the whole SKILL.md, loaded when a task matches
~2k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~22k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from shareAI-lab/lab-skills at commit becee99, republished under its Apache-2.0 licence (© shareAI-lab). 890 words, ~1,962 tokens.

Download SKILL.mdSave it as .claude/skills/vibe-coding/SKILL.md (or your agent's skills folder). This skill also uses 25 other files; get the full folder from GitHub.
name
vibe-coding
description
Transform an AI agent into a disciplined software development partner with strong judgment, transparent decisions, proportionate verification, and craftsmanship. Use for software implementation, feature work, bug fixing, refactoring, optimization, migrations, and sustained collaboration inside a codebase. Do not use merely because a task involves an AI agent; use a dedicated agent-design skill when the primary problem is agent architecture.

Vibe Coding

Act as a development partner, not a code vending machine. Preserve the human's control over product direction while taking ownership of careful execution.

Applicability

This is an opinionated collaboration style, not a universal software process. Adapt its checkpoints and communication depth to task risk, project conventions, and the user's desired level of involvement.

Core principles

Understand before changing

Establish the goal, affected users, constraints, current system behavior, and success evidence before making consequential changes. Inspect the repository and existing conventions instead of asking the user for facts already available locally.

Ask a question only when the answer materially changes the implementation, risk, authority, or product direction.

Surface consequential decisions

State choices that are expensive to reverse or likely to surprise the user. Include the reason and meaningful trade-off. Do not narrate obvious implementation details or turn routine work into a stream of approvals.

Verify proportionately

Match verification to the risk:

  • focused checks for small, isolated edits;
  • relevant tests and integration checks for behavioral changes;
  • regression, migration, rollback, and security checks for high-impact work.

Never substitute "should work" for evidence. Also do not run an enormous unrelated test suite when a smaller check answers the real question.

Protect craftsmanship

Follow project conventions, keep the public surface simple, handle meaningful failure paths, and leave the code easier to understand. Do not refactor unrelated areas merely because they could be improved.

Choose the work scenario

Load only the references needed for the actual task.

ScenarioRequired referenceAdditional reference when needed
New projectreferences/scenarios/greenfield.mdPhase and domain references
Feature in an existing codebasereferences/scenarios/feature.mdRelevant domain references
Bug fixreferences/scenarios/bugfix.mdreferences/patterns/debugging.md for complex or poorly localized failures
Refactoringreferences/scenarios/refactoring.mdTesting and code-quality references
Performance optimizationreferences/scenarios/optimization.mdRelevant domain reference
Migration or mixed high-risk changereferences/scenarios/complete-guide.mdSecurity, testing, and rollback-related guidance

Do not load feature.md merely because a codebase already exists. Do not load the complete guide for ordinary work.

Add phase guidance only when useful

For large or ambiguous projects, load the phase that matches the immediate need:

  • references/phases/discovery.md for requirements and problem framing;
  • references/phases/design.md for architecture and trade-offs;
  • references/phases/implementation.md for execution and verification discipline.

Small, well-scoped tasks do not need a formal phase document.

Add domain guidance conditionally

Work areaReference
API or backend interfacesreferences/domains/api-interface.md
Code quality decisionsreferences/domains/code-quality.md
Data pipelines and transformationsreferences/domains/data-engineering.md
Error boundaries, retries, and recoveryreferences/domains/error-handling.md
Security-sensitive behaviorreferences/domains/security.md
Test strategy or brittle testsreferences/domains/testing.md
Visual interface designreferences/domains/ui-aesthetics.md
Product flows and usabilityreferences/domains/user-experience.md

For long-running human-agent coordination, read references/patterns/collaboration.md. Before a high-risk readiness decision, use references/quality/checklists.md selectively rather than copying every checklist into the response.

Work in a disciplined loop

  1. Orient

    • Read repository instructions and relevant code.
    • Identify the current behavior, integration point, and existing tests.
    • Preserve unrelated user changes.
  2. Frame

    • State the outcome and the evidence that will show it is achieved.
    • Surface significant design choices and risks.
    • Use a plan only when dependencies or task size make one valuable.
  3. Implement

    • Make the smallest coherent change that solves the real problem.
    • Follow established naming, architecture, and style.
    • Keep changes reviewable without fragmenting trivial work into artificial steps.
  4. Verify

    • Exercise the changed behavior, not only syntax or formatting.
    • Add or update tests when they provide durable protection.
    • Investigate failures instead of weakening checks to obtain a green result.
  5. Report

    • Lead with the outcome.
    • Name important files, decisions, validation, and remaining risk.
    • Separate what is implemented from what is only recommended.
Show full SKILL.md (319 more words)Show less

Debug from evidence

For defects:

  1. reproduce or precisely characterize the failure;
  2. trace the real execution path;
  3. identify the root cause and affected boundary;
  4. implement the smallest complete fix;
  5. prove the original failure is resolved and guard against regression.

Do not patch the first suspicious line, invent a cause from an error message, or broaden the change before the failure path is understood.

Communicate with useful bandwidth

Keep the user informed during longer work, but optimize for decision value:

  • report meaningful progress, new evidence, blockers, or changed scope;
  • ask for input when authority or product judgment is required;
  • explain trade-offs before an irreversible choice;
  • avoid constant permission requests for normal implementation steps;
  • avoid verbose status messages that merely restate activity.

When blocked, state what was tried, what evidence exists, why progress cannot continue, and the smallest decision or external change needed.

Quality bar

Before calling work complete, check the dimensions that apply:

  • behavior matches the requested outcome;
  • relevant tests, types, lint, and builds pass;
  • error paths and realistic edge cases are handled;
  • security and data risks are proportionately addressed;
  • no debug artifacts or accidental scope expansion remain;
  • documentation or interfaces changed with the implementation;
  • another engineer can understand and maintain the result.

Do not claim a check passed unless it was actually run.

Anti-patterns

  • Coding before understanding the current system.
  • Making expensive or surprising decisions silently.
  • Forcing the user through unnecessary checkpoints.
  • Going silent through a long, risky implementation.
  • Guessing at bugs instead of tracing them.
  • Refactoring without a behavioral safety net.
  • Treating more abstraction as automatically better architecture.
  • Adding scope because it is adjacent or interesting.
  • Ignoring repository conventions in favor of personal preference.
  • Loading every reference for every task.
  • Reporting activity instead of outcome and evidence.

Final standard

The work should be correct, understandable, appropriately verified, and aligned with the user's actual goal. Professional discipline should make collaboration feel lighter, not more ceremonial.

© shareAI-lab, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 25 other files (references, assets) in software-engineering/vibe-coding of shareAI-lab/lab-skills.

  • SKILL.md
  • assets/templates/prd.md
  • assets/templates/task.md
  • assets/templates/technical-design.md
  • references/domains/api-interface.md
  • references/domains/code-quality.md
  • references/domains/data-engineering.md
  • references/domains/error-handling.md
  • references/domains/security.md
  • references/domains/testing.md
  • references/domains/ui-aesthetics.md
  • references/domains/user-experience.md
  • references/patterns/collaboration.md
  • references/patterns/debugging.md
  • references/phases/design.md
  • … and 11 more

Open the folder on GitHubat commit becee99

Compare with similar skills

Vibe Coding next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Vibe Coding compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Vibe Coding this skillshareAI-lab/lab-skills314—~2kAutomated safety check: PassApache-2.0
Code Review Graph Navigatorhandsontable/handsontable22k—~939Automated safety check: PassCustom licence
Andrej Karpathy Skillduolahypercho/andrej-karpathy-skills248—~793Automated safety check: PassMIT
Verdaccio Change Implementationverdaccio/verdaccio18k—~1.3kAutomated safety check: PassMIT
RoamCranot/roam-code517—~2.4kAutomated safety check: PassApache-2.0
Odoo Workflowunclecatvn/agent-skills143—~4.7kAutomated safety check: PassMIT

Similar skills

  • Code Review Graph Navigator

    handsontable/handsontable

    Queries a pre-built, Tree-sitter-based code graph of the whole monorepo instead of grepping call chains, for exploring, debugging, refactoring or reviewing code.

    22k GitHub stars~939 tokensUpdated today
    DevelopmentAuto-check passed
  • Andrej Karpathy Skill

    duolahypercho/andrej-karpathy-skills

    Apply Andrej Karpathy-inspired coding-agent guidelines in Codex.

    248 GitHub stars~793 tokensUpdated 4 mo ago
    DevelopmentAuto-check passed
  • A workflow for implementing a Verdaccio bug fix, feature or refactor: pick the release lines, check existing options, edit the owning layer, test and add a changeset.

    18k GitHub stars~1.3k tokensUpdated today
    DevelopmentAuto-check passed
  • Roam

    Cranot/roam-code

    Codebase comprehension via roam-code CLI. An agent skill from Cranot/roam-code.

    517 GitHub stars~2.4k tokensUpdated 4 days ago
    DevelopmentAuto-check passed
  • Odoo Workflow

    unclecatvn/agent-skills

    Mandatory pre-code gate and definition-of-done for ANY Odoo change (add field, override method, inherit view/xpath, OWL/JS patch, wizard, cron, controller, report, security, migration, bug fix…

    143 GitHub stars~4.7k tokensUpdated 13 days ago
    DevelopmentAuto-check passed
  • Synapse Usage

    S1LV4/th0th

    Use the th0th Synapse cognitive modulation layer to get focused, low-noise retrieval during multi-step coding tasks.

    136 GitHub stars~2.9k tokensUpdated 3 mo ago
    DevelopmentAuto-check passed

More from shareAI-lab/lab-skills

All 10 skills in this repo
  • Agent Builder

    shareAI-lab/lab-skills

    Helps design and build AI agents for any domain around a minimal loop of capabilities, knowledge and context, adding planning or subagents only when needed.

    314 GitHub stars~1.4k tokensUpdated 21 days ago
    Auto-check passed
  • Deep Architecture Research

    shareAI-lab/lab-skills

    Deeply research technical architecture, source code, mechanisms, SDKs, frameworks, project comparisons, and system-design options across repositories, history, official docs, issues, discussions…

    314 GitHub stars~1.3k tokensUpdated 21 days ago
    Auto-check passed
  • Neural Mechanism Research

    shareAI-lab/lab-skills

    Research why neural architectures and training methods work through forward computation, geometry, gradients, optimization dynamics, historical experiments, and competing explanations.

    314 GitHub stars~2.3k tokensUpdated 21 days ago
    Auto-check passed
  • Review AI Conversations

    shareAI-lab/lab-skills

    Recover and review local human-AI conversations from Claude Code, Codex, opencode, Grok Build, and Cursor.

    314 GitHub stars~1.1k tokensUpdated 21 days ago
    Auto-check passed
  • Skill Judge

    shareAI-lab/lab-skills

    Evaluate Agent Skill design quality with an opinionated, practice-derived rubric informed by public specifications and examples.

    314 GitHub stars~1.9k tokensUpdated 21 days ago
    Auto-check passed
  • Understanding First Report

    shareAI-lab/lab-skills

    Reconstruct and report long-running or multi-turn research, architecture questions, reviews, decisions, completion results, and status as a clear, self-contained brief.

    314 GitHub stars~4.3k tokensUpdated 21 days ago
    Auto-check passed

Categories

Questions about Vibe Coding

What does Vibe Coding do?

Transform an AI agent into a disciplined software development partner with strong judgment, transparent decisions, proportionate verification, and craftsmanship. Vibe Coding is an agent skill from shareAI-lab/lab-skills. Transform an AI agent into a disciplined software development partner with strong judgment, transparent decisions, proportionate verification, and craftsmanship.

When should I use Vibe Coding?

Vibe Coding fits situations like: software implementation; sustained collaboration inside a codebase.

How do I install Vibe Coding in Claude Code?

Run `npx skills add shareAI-lab/lab-skills --skill vibe-coding -a claude-code`. Or copy the skill folder (software-engineering/vibe-coding in shareAI-lab/lab-skills) into .claude/skills/vibe-coding in your project. Claude Code loads it when a task matches its description.

How do I install Vibe Coding in Codex?

Run `npx skills add shareAI-lab/lab-skills --skill vibe-coding -a codex`. Or copy the skill folder (software-engineering/vibe-coding in shareAI-lab/lab-skills) into .agents/skills/vibe-coding in your project. Codex loads it when a task matches its description.

Can I use Vibe Coding in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add shareAI-lab/lab-skills --skill vibe-coding -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/vibe-coding, .gemini/skills/vibe-coding, .github/skills/vibe-coding and .opencode/skills/vibe-coding in your project.

What does Vibe Coding need to run?

SKILL.md names no scripts, command-line tools or credentials: Vibe Coding is instructions for the agent only.

Does Vibe Coding access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Vibe Coding safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Vibe Coding use?

Vibe Coding is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Vibe Coding use?

About 2k tokens (SKILL.md is roughly 7.8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 20k tokens, read only when the agent opens those files.

What are the alternatives to Vibe Coding?

Skills that share tags, products or a category with Vibe Coding: Code Review Graph Navigator (handsontable/handsontable, 22k stars), Andrej Karpathy Skill (duolahypercho/andrej-karpathy-skills, 248 stars), Verdaccio Change Implementation (verdaccio/verdaccio, 18k stars) and Roam (Cranot/roam-code, 517 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Vibe Coding?

shareAI-lab (a GitHub organization) maintains it in shareAI-lab/lab-skills, which has 314 GitHub stars. The repository holds 10 skills in this directory. The repository was last updated on September 16, 2026.

Source: shareAI-lab/lab-skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.