Website & GitHub Blueprint for AI Agents avatar

Website & GitHub Blueprint for AI Agents

Pricing

from $0.15 / quick blueprint

Go to Apify Store
Website & GitHub Blueprint for AI Agents

Website & GitHub Blueprint for AI Agents

Turn a public GitHub, GitLab, Bitbucket repo or website into an evidence-backed blueprint for Cursor, Claude Code, or Codex. Quick needs no AI key. Deep uses your key and returns an Agent Launch Pack plus a portable Build Skill.

Pricing

from $0.15 / quick blueprint

Rating

0.0

(0)

Developer

Viktor Kondas

Viktor Kondas

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

3 days ago

Last modified

Share

Turn a public GitHub, GitLab, Bitbucket repo or website into an evidence-backed blueprint for Cursor, Claude Code, or Codex.

This is not a GitHub scraper. It is a reconstruction pack: Analysis Confidence, Source Coverage, and (in Deep mode) an Agent Launch Pack plus a portable Build Skill.

Store: apify.com/kondasviktor/website-github-blueprint

Pricing (PPE)

EventWhenPrice
Actor startPlatform defaultPlatform default
quick-blueprintOne Quick Blueprint (any target)$0.15
deep-repository-blueprintDeep pack for one public repository$1.25
deep-website-blueprintDeep pack for one public website (+ optional screenshots)$1.50

Minimum charge $0.15. Prefer PPE only (not + usage). Deep mode uses your Gemini / OpenAI / Anthropic / OpenRouter key — model spend is billed by your provider, not Apify.

Failed runs are not charged.

What it does

  • Public GitHub / GitLab / Bitbucket repositories and public HTTPS websites
  • Quick Blueprint — deterministic map + Source Coverage, no AI key
  • Deep Blueprint — BYOK architecture, journeys, implementation plan, acceptance criteria, Agent Launch Pack ZIP, Build Skill ZIP
  • Build-brief controls: goal, preferred stack, focus areas, primary coding agent
  • Optional repository API token for rate limits only (never unlocks private repos)
  • Self-contained dataset record for API / automation clients

Limitations (v1)

  • Public targets only — private repos, login walls, CAPTCHA, and paywalls are rejected
  • Website crawl capped at 1–10 same-origin pages (default 5)
  • Repository file budgets are bounded (Quick ~50 files, Deep ~200)
  • Deep mode requires a caller-supplied AI key
  • Screenshots are Deep-website only and cover the first page (desktop + mobile)
  • Not a full clone, not a GitHub stars/forks scraper, not a SaaS dashboard

Quick start — GitHub repo (no AI key)

{
"url": "https://github.com/molin-ai/skills",
"depth": "quick"
}

Quick start — public website (no AI key)

{
"url": "https://molin.ai/",
"depth": "quick",
"maxPages": 5
}

Deep Blueprint — add your key

{
"url": "https://molin.ai/",
"depth": "deep",
"aiProvider": "gemini",
"apiKey": "set-this-only-in-Apify-secret-input",
"buildGoal": "mvp",
"preferredStack": "Next.js, PostgreSQL, and Tailwind CSS",
"focusAreas": ["frontend", "integrations", "testing"],
"agentTarget": "cursor",
"maxPages": 5,
"includeScreenshots": true
}

Output

  • Dataset: one row with the full validated Blueprint, confidence, stack, and artifact manifest
  • Key-value store:
    • OUTPUT — same structured record
    • ARTIFACT_MANIFEST — keys that actually exist for this run
    • QUICK_BLUEPRINT or BUILD_BLUEPRINT — Markdown
    • SOURCE_COVERAGE — evidence coverage
    • Deep only: AGENT_LAUNCH_PACK and BUILD_SKILL (ZIP)
    • Optional: SCREENSHOT_* WebP captures

Examples

Use with Claude, Cursor, and MCP

Available on the Apify MCP server. Pass a public repo or site URL with depth: "quick" to try without an AI key. For Deep packs, pass your provider key as a secret input.

Analyze only public content you have rights to inspect. Respect each site's terms and robots rules. You are responsible for how you use the results. Not legal advice.

FAQ

Is this a GitHub scraper?
No. It produces an implementation blueprint and agent pack, not stars/forks/metadata dumps.

Do I need an AI key for Quick mode?
No. Quick is deterministic. Deep requires your Gemini, OpenAI, Anthropic, or OpenRouter key.

Can I analyze private repositories?
No. Private and auth-walled targets are rejected. An optional repository token only raises public API rate limits.

What is the Agent Launch Pack?
A ZIP of portable prompts and context files tailored for Cursor, Claude Code, Codex, or a generic agent.

All Actors

Feedback

Contact · Built by Vibe Coder's Life