{
  "markdown": "# Cairo — Pricing Diagnose\n\n> A Claude skill that self-diagnoses the pricing system of your AI product.\n>\n> Run `/pricing-diagnose` inside your project. Get a Reality Check Report with file paths, line numbers, estimated impact, and fix effort.\n\n---\n\n## Quick prompt-only check\n\nWant to try Cairo before installing the skill? Open Claude Code at your project root and paste this:\n\n```text\nAct as Cairo, a pricing-system diagnostic reviewer for AI products.\n\nQuickly evaluate this codebase's pricing system. Do not make code changes.\n\n1. Detect whether the pricing model is tiered, usage-based, hybrid, or unknown.\n2. Inspect only the highest-signal pricing files first: pricing page/components, plan config, billing/checkout code, Stripe/Paddle integration, credit or usage ledger, lifecycle/cancel flow, and relevant git history.\n3. Score these 6 signals briefly:\n   - Page-code consistency\n   - Experimentation velocity\n   - Hidden SKUs + experimentation infrastructure\n   - Billing stack maturity\n   - Unit clarity\n   - Lifecycle coverage\n4. If the model is usage-based or hybrid, also check:\n   - Credit carryover\n   - Decoy pricing\n   - One-time credit purchase\n   - Cancellation save offer\n   - Credit meter\n   - Credit usage email\n5. Output a compact Reality Check:\n   - Pricing model + confidence\n   - Diagnostic map\n   - 6-signal score table\n   - Top 3 findings with exact file:line, impact, and estimated effort\n\nKeep it concise. Prefer concrete file evidence over broad advice.\n```\n\nFor the full repeatable workflow, install the plugin or skill below and run `/pricing-diagnose`.\n\n---\n\n## What it does\n\nCairo reads your codebase, git history, and config files — then scores your pricing maturity across **6 signals** and (for usage-based products) **6 deep checks**.\n\nIt detects whether your pricing model is:\n- **tiered** — fixed plans (Free / Pro / Enterprise)\n- **usage-based** — credits, tokens, metered billing\n- **hybrid** — both (most AI products)\n\n…and weights the signals + runs the right deep checks accordingly.\n\n### The 6 signals\n\n| # | Signal | What it measures |\n|---|---|---|\n| 1 | Page-Code Consistency | What your pricing page advertises vs. what your code actually charges |\n| 2 | Experimentation Velocity | How often pricing changes ship (git history) |\n| 3 | Hidden SKUs + Experimentation Infrastructure | Plans that exist in code but aren't shown; A/B SDK presence |\n| 4 | Billing Stack Maturity | Stripe Meters? Custom ledger? Idempotent webhooks? |\n| 5 | Unit Clarity | Does the user understand what 1 credit / 1 unit costs them? |\n| 6 | Lifecycle Code Coverage | Exposure → Signup → Usage → Expansion → Churn → Recovery |\n\n### The 6 usage-based deep checks (run for usage-based / hybrid)\n\n| # | Check | Why it matters |\n|---|---|---|\n| U1 | Credit Carryover | Hard month-end reset is the #1 user complaint |\n| U2 | Decoy Pricing | No anchor pack → conversion left on the table |\n| U3 | One-Time Credit Purchase | Missing top-up forces unwanted subscription churn |\n| U4 | Cancellation Flow (bonus credits vs % discount) | No save offer = 100% churn |\n| U5 | Credit Meter | No live balance = \"credits disappeared\" support tickets |\n| U6 | Credit Usage Email | No threshold alerts = surprise depletion churn |\n\n---\n\n## Install\n\n### Option A — Plugin (recommended)\n\nThis repo doubles as its own Claude Code plugin marketplace. Add it once, then install:\n\n```\n/plugin marketplace add getklaim/cairo\n/plugin install cairo@klaim\n```\n\nTo run the plugin from a local checkout (e.g. for development):\n\n```bash\ngit clone https://github.com/getklaim/cairo.git\nclaude --plugin-dir ./cairo\n```\n\n**Invocation when installed as a plugin:**\n\n```\n/cairo:pricing-diagnose\n```\n\n### Option B — Manual single-file install\n\nCairo's skill works standalone — just drop the `SKILL.md` into a Claude skills directory.\n\n```bash\n# Project-scoped\nmkdir -p .claude/skills/pricing-diagnose\ncurl -L https://raw.githubusercontent.com/getklaim/cairo/main/skills/pricing-diagnose/SKILL.md \\\n  -o .claude/skills/pricing-diagnose/SKILL.md\n\n# Or user-scoped (all projects)\nmkdir -p ~/.claude/skills/pricing-diagnose\ncurl -L https://raw.githubusercontent.com/getklaim/cairo/main/skills/pricing-diagnose/SKILL.md \\\n  -o ~/.claude/skills/pricing-diagnose/SKILL.md\n```\n\n**Invocation when installed manually:**\n\n```\n/pricing-diagnose\n```\n\nReload Claude Code after installing either way.\n\n---\n\n## Use\n\nFrom the root of any project. Replace `/pricing-diagnose` with `/cairo:pricing-diagnose` if you installed via the plugin path.\n\n```bash\n# Full diagnosis (auto-detects pricing model)\n/pricing-diagnose\n\n# Fast first-pass diagnosis\n/pricing-diagnose --quick\n\n# Focus a single signal\n/pricing-diagnose --signal 5\n\n# Force pricing model when auto-detection is wrong\n/pricing-diagnose --pricing-model usage-based\n\n# Run only the 6 usage-based deep checks\n/pricing-diagnose --usage-deep\n\n# Combined analysis across split repos (marketing site + billing app)\n/pricing-diagnose --related-repos ../billing-app\n```\n\nYou'll get a Reality Check Report with:\n- Pricing model + diagnostic map\n- Score per signal (with the file path that drove the score)\n- Top 3 priority findings with exact file locations and impact estimates\n- Estimated effort per finding\n\nSee [skills/pricing-diagnose/SKILL.md](./skills/pricing-diagnose/SKILL.md) for the full spec.\n\n---\n\n## Privacy & security\n\n- **Runs entirely on your machine.** Cairo analyzes files locally — no code, no findings, no scores leave your environment.\n- **No telemetry.** Cairo contains no analytics SDKs and makes no outbound network calls.\n- **Read-only.** Cairo never modifies your code. It produces a report.\n\nFull details: [PRIVACY.md](./PRIVACY.md)\n\n---\n\n## What Cairo will *not* tell you\n\nHonest limitations (also in `SKILL.md`):\n\n- Short git history → Signal 2 (experimentation velocity) becomes unreliable\n- Custom in-house billing systems weaken Signal 4 inference\n- Decoy pricing detection (U2) without telemetry can only flag *possible* decoys\n- Impact estimates (\"~30% of support tickets\") are heuristics from code signals — not statistically calibrated\n- Cannot infer intent (deliberate ambiguity vs. inexperience look the same in code)\n\nThe report explicitly marks sections where these limitations apply.\n\n---\n\n## How does this compare to…\n\n| Tool | Approach | Cairo difference |\n|---|---|---|\n| Tierly | External page scoring (11 axes) | Cairo reads your *code*, not just your page |\n| LLM Price Check | Token unit-price comparison | Cairo evaluates how *your product* exposes cost |\n| Helicone / Langfuse | LLM observability | Cairo diagnoses the pricing structure itself |\n| ProfitWell / Baremetrics | Subscription metrics | Cairo doesn't need historical revenue data — it works on day 1 |\n\nExternal tools observe from outside. Cairo runs *inside* your project.\n\n---\n\n## License\n\nMIT. See [LICENSE](./LICENSE).\n\n---\n\n## Built by Klaim\n\nCairo is open-sourced by [Klaim](https://getklaim.com), a pricing experimentation platform for AI products.\n\nIf your Cairo report flags issues across Signals 1, 2, 3, and 5 and you'd rather buy than build — Klaim covers those signals as a product. Cairo will keep working without it; the 90% of value is the self-diagnosis itself.\n\nIssues, ideas, PRs welcome.\n",
  "bytes": 7246,
  "sha": "1947c685956f96134bdf532026da17f0ee2f3dd30b31fdaf82d258b42ffde0f6",
  "repo_slug": "getklaim/cairo",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_getklaim_cairo_cairo_134ecd80/readme"
}