New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Genesis

by ayush488-glitch in ayush488-glitch/genesis-kit

Start a specification-first project, adopt or resume a repository, and operate the Genesis repository-native human-AI development harness.

MITAuto-check passed

Install Genesis

skills CLI
$ npx skills add ayush488-glitch/genesis-kit --skill genesis -a claude-code

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

GitHub CLI
$ gh skill install ayush488-glitch/genesis-kit genesis --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/ayush488-glitch/genesis-kit.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/genesis .claude/skills/genesis && 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
genesis
GitHub stars
152
Token cost
~1.9k tokens
SKILL.md length
1,036 words
Files
1
Skills in repo
1
Repo updated
First seen
Licence
MIT

At a glance

Start a specification-first project, adopt or resume a repository, and operate the Genesis repository-native human-AI development harness.

  • Works in 8 steps: Interview the human about users,… → Record consequential assumptions,… → Give every functional, non-functional,… → …
  • SKILL.md covers Required policy, Entry points, Session contract and New-product workflow, plus 2 more sections
  • Calls node

What it does

Genesis is an agent skill from ayush488-glitch/genesis-kit. Start a specification-first project, adopt or resume a repository, and operate the Genesis repository-native human-AI development harness.

Its SKILL.md is about 1.9k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

The repository describes itself as: Turn any empty repo into a loop-ready repo: a portable per-project spine + anti-slop build loops for AI coding agents. The licence is MIT.

Example prompts

  • “/genesis”

Requirements

  • Python 3

Workflow steps

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

  1. Interview the human about users, outcomes, constraints, non-goals, trust boundaries, risks, success criteria, and unknowns.
  2. Record consequential assumptions, decisions, and sourced knowledge through genesis record while filling SPEC.md.
  3. Give every functional, non-functional, and acceptance requirement a stable FR-*, NFR-*, or AC-* identifier.
  4. Run genesis spec check . Show the specification and stop for explicit human approval; never approve it on the human's behalf.
  5. After approval, record it with genesis spec approve --human NAME --reason "...".
  6. Create bounded tasks using approved --requirement identifiers and at least one executable --gate per task.
  7. Run genesis plan check , show generated .genesis/PLAN.md, and stop for explicit human approval.
  8. Record approval with genesis plan approve --human NAME --reason "...", then implement only the activated task.

What it can do on your machine

Read from SKILL.md and the folder at commit c5afb26. 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

    Shell commands in SKILL.md call:

    • node

    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

Genesis loads about 1.9k tokens when it runs. Until then it costs about 37 tokens; SKILL.md has 1,036 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~37
When it runs · the whole SKILL.md, loaded when a task matches
~1.9k

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 ayush488-glitch/genesis-kit at commit c5afb26, republished under its MIT licence (© ayush488-glitch). 1,036 words, ~1,945 tokens.

Download SKILL.mdSave it as .claude/skills/genesis/SKILL.md (or your agent's skills folder).
name
genesis
description
Start a specification-first project, adopt or resume a repository, and operate the Genesis repository-native human-AI development harness.

Genesis

Use Genesis when starting a project, adopting an existing repository, resuming work from a cold session, proving a bounded task, or inspecting its software-factory state.

Required policy

Load the official Ponytail skill at full before architecture or code changes. Do not replace it with a copied summary.

Entry points

  • New product: genesis init <repo> --workflow new-product --objective "...". Remain in discovery/specification until human approval.
  • New task-only repository: genesis init <repo> --objective "...".
  • Existing repository: genesis adopt <repo> first; run again with --write only after the human accepts the discovery report.
  • Existing legacy spine: genesis migrate <repo> first, then --write. Migration preserves every legacy file.
  • Resume: read .genesis/KICKOFF.md, then run genesis brief <repo> for the current phase and task contract.
  • Connect cold coding-agent sessions: inspect genesis agent connect <repo> first, then use --write only with human approval.
  • Ask the index: genesis query <repo> search|scope|defines|callers|callees|impact|neighbours|path (--json to parse). Expose the same questions to an MCP host with genesis mcp <repo>.
  • See the codebase: genesis serve <repo> --open draws it as a live map, reindexes on save and is read-only.

If genesis is not on PATH, run node <genesis-kit>/tools/genesis.mjs.

Session contract

Before editing, report state → evidence → blocker → next action. Inspect only the decisions, proof and code the active task needs.

Query the index before reading widely or grepping. genesis query <repo> search NAME finds where a name is defined; scope PATH gives a file or directory's dependencies in both directions; callers/callees give call edges; impact PATH gives everything that transitively imports a file, which is the blast radius to check before editing shared code; path FROM TO shows how two files connect. Search the repository before adding anything, and query it before searching.

Index answers are advisory static analysis, not authority: confirm in source before relying on one. A result marked ambiguous means several definitions matched and none were ruled out, so read the candidates rather than taking the first. Symbols are found by an anchored line scan and only JavaScript, TypeScript and Python are extracted, so absence from the index is not evidence of absence in the repository.

Obey the workflow phase instruction in KICKOFF.md. During discovery, specification, and planning, do not write product implementation code.

New-product workflow

  1. Interview the human about users, outcomes, constraints, non-goals, trust boundaries, risks, success criteria, and unknowns.
  2. Record consequential assumptions, decisions, and sourced knowledge through genesis record while filling SPEC.md.
  3. Give every functional, non-functional, and acceptance requirement a stable FR-*, NFR-*, or AC-* identifier.
  4. Run genesis spec check <repo>. Show the specification and stop for explicit human approval; never approve it on the human's behalf.
  5. After approval, record it with genesis spec approve <repo> --human NAME --reason "...".
  6. Create bounded tasks using approved --requirement identifiers and at least one executable --gate per task.
  7. Run genesis plan check <repo>, show generated .genesis/PLAN.md, and stop for explicit human approval.
  8. Record approval with genesis plan approve <repo> --human NAME --reason "...", then implement only the activated task.

Create one bounded task. Every mandatory check must pass against the current source hash; pending, skipped, failed, missing, and stale proof block completion. Medium risk and above requires independent review.

Use genesis record knowledge|decision|assumption|invariant for durable context and provenance. Do not leave a binding decision or discovered constraint only in chat.

Decision and knowledge records require --title and --text; assumptions and invariants require --text. Add --source whenever provenance is known.

Non-low-risk tasks automatically receive an independent-review gate. A separate human approves it only after checking the diff and proof:

bash
genesis control approve <repo> <task-id> --gate independent-review --human <name> --reason "..."

Checkpoint before handing off:

bash
genesis checkpoint <repo>

The generated kickoff and dashboard are read-only projections of project.json. Raw traces stay local and secrets are redacted. Propose cleanup and learned rules; never delete code or promote harness rules silently.

Show full SKILL.md (419 more words)Show less

Verified autonomy

Use genesis context <repo> [task-id] to retrieve bounded task context; --id retrieves a full durable record. Corrections use record ... --supersedes <id> so obsolete knowledge stays out of active context without erasing history.

For autonomous execution, declare file scope and outcome scenarios (actor, scope, environment, duration, observable and executable gate). Record only real user authorization with genesis authorize grant; then genesis run invokes the bounded host command and verifies completion. Existing authorization persists across sessions. Read the kit's docs/autonomy-contract.md for argument and receipt schemas.

Never route completion through task set. Run executable gates; runtime gates must independently check their target and return a source-bound JSON receipt. Medium/high-risk work still requires independent review of current evidence. Do not impersonate a reviewer. Resume a stopped review with genesis run; unchanged successful worker attempts are reused. After a crash, use genesis recover, inspect uncertain side effects and resume explicitly. Do not automatically replay side effects.

Keep causal incident hypotheses separate from supported diagnoses. Learning proposals require baseline/candidate evaluation with development and holdout cases, a matching policy artifact, independent review, explicit promotion and rollback. Candidate work directories are not sandboxes; host permissions must protect the evaluator and holdout. Never treat self-reported success or a tied evaluation as measured improvement.

Context economy

Use one bounded brief for the active slice; do not load all project.json, research history or prior tool transcripts by default. The compact packet retains binding invariants and active rules. Fetch full records with context --id when a summary is insufficient, and respect an explicit budget error. --full restores full optional record bodies with a suitable --bytes budget.

Use --since only after receiving and retaining the complete packet for that fingerprint. The kickoff's fingerprint alone is not that packet. Token estimates are labeled heuristics; measure actual host usage before claiming savings. Use the original recipes for research, plan, implement, verify and recover without requiring a specific model. Run genesis dashboard <repo> --open for the read-only control panel; its buttons copy commands, never execute or approve them. Run genesis serve <repo> --open for the live panel: it watches sources, reindexes incrementally on save, and draws the repository as a symbol map with call edges. Both are projections; neither approves work. Run genesis index <repo> if the index is stale and nothing is watching.

The brief already carries scope cards and a symptom map read from the index: what each declared scope depends on and what depends on it, and the declarations named by the task text. Start there rather than re-deriving them.

© ayush488-glitch, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/genesis of ayush488-glitch/genesis-kit.

Open the folder on GitHubat commit c5afb26

Compare with similar skills

Genesis 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.

Genesis compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Genesis this skillayush488-glitch/genesis-kit152—~1.9kAutomated safety check: PassMIT
Resume Tailorreactive-resume/reactive-resume44k—~8.3kAutomated safety check: PassMIT
Resume Bullet Writerreactive-resume/reactive-resume44k—~8kAutomated safety check: PassMIT
Reactive Resume Builderreactive-resume/reactive-resume44k—~2kAutomated safety check: PassMIT
Resume Content Guidereactive-resume/reactive-resume44k—~11kAutomated safety check: WarnMIT
AdoptDonchitos/Claude-Code-Game-Studios26k—~1.2kAutomated safety check: NotesMIT

Similar skills

  • Resume Tailor

    reactive-resume/reactive-resume

    Tailors one resume to one specific job posting. An agent skill from reactive-resume/reactive-resume.

    44k GitHub stars~8.3k tokensUpdated yesterday
    Business, Finance & HRAuto-check passed

  • Resume Bullet Writer

    reactive-resume/reactive-resume

    Turns resume duties into truthful impact bullets. An agent skill from reactive-resume/reactive-resume.

    44k GitHub stars~8k tokensUpdated yesterday
    Business, Finance & HRAuto-check passed

  • Reactive Resume Builder

    reactive-resume/reactive-resume

    Builds resumes as valid JSON for the open-source Reactive Resume app by interviewing you, and can track job applications through its MCP tools.

    44k GitHub stars~2k tokensUpdated yesterday
    Business, Finance & HRAuto-check passed

  • Resume Content Guide

    reactive-resume/reactive-resume

    Decides what information belongs on a resume or CV, what to leave off and in what order, and audits an existing resume against those norms.

    44k GitHub stars~11k tokensUpdated yesterday
    Business, Finance & HRAuto-check: warnings

  • Adopt

    Donchitos/Claude-Code-Game-Studios

    Brownfield audit — do existing artifacts actually work?. An agent skill from Donchitos/Claude-Code-Game-Studios.

    26k GitHub stars~1.2k tokensUpdated 3 days ago
    Game DevelopmentAuto-check: notes

  • Agent Specification

    ruvnet/ruflo

    Agent skill for specification - invoke with $agent-specification

    74k GitHub starsUsed in 2 repos~1.8k tokens
    Product & Project ManagementAuto-check passed

Questions about Genesis

What does Genesis do?

Start a specification-first project, adopt or resume a repository, and operate the Genesis repository-native human-AI development harness. Genesis is an agent skill from ayush488-glitch/genesis-kit. Start a specification-first project, adopt or resume a repository, and operate the Genesis repository-native human-AI development harness.

How do I install Genesis in Claude Code?

Run `npx skills add ayush488-glitch/genesis-kit --skill genesis -a claude-code`. Or copy the skill folder (skills/genesis in ayush488-glitch/genesis-kit) into .claude/skills/genesis in your project. Claude Code loads it when a task matches its description.

How do I install Genesis in Codex?

Run `npx skills add ayush488-glitch/genesis-kit --skill genesis -a codex`. Or copy the skill folder (skills/genesis in ayush488-glitch/genesis-kit) into .agents/skills/genesis in your project. Codex loads it when a task matches its description.

Can I use Genesis 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 ayush488-glitch/genesis-kit --skill genesis -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/genesis, .gemini/skills/genesis, .github/skills/genesis and .opencode/skills/genesis in your project.

What does Genesis need to run?

Going by SKILL.md and its folder, Genesis needs the command-line tools its instructions call (node). Our summary lists: Python 3.

Does Genesis 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 Genesis 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 Genesis use?

Genesis is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Genesis use?

About 1.9k 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.

What are the alternatives to Genesis?

Skills that share tags, products or a category with Genesis: Resume Tailor (reactive-resume/reactive-resume, 44k stars), Resume Bullet Writer (reactive-resume/reactive-resume, 44k stars), Reactive Resume Builder (reactive-resume/reactive-resume, 44k stars) and Resume Content Guide (reactive-resume/reactive-resume, 44k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Genesis?

ayush488-glitch (a GitHub user) maintains it in ayush488-glitch/genesis-kit, which has 152 GitHub stars. The repository was last updated on September 28, 2026.

Source: ayush488-glitch/genesis-kit on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.