block · cnghockey · Apr 9, 2026
diff --git a/bitcoin-ai-tools/SKILL.md b/bitcoin-ai-tools/SKILL.md
@@ -0,0 +1,144 @@
+---
+name: bitcoin-ai-tools
+description: Access 33+ AI services (image, video, audio, OCR, translation, communication) by paying per request with Bitcoin Lightning — no API keys or signups required.
+author: cnghockey
+version: "1.0"
+tags:
+  - ai
+  - bitcoin
+  - lightning
+  - payments
+  - image-generation
+  - audio
+  - video
+  - no-signup
+---
+
+# Bitcoin AI Tools (Sats4AI)
+
+## Purpose
+
+Use Sats4AI to call specialized AI services — image generation, background removal, TTS, transcription, video, OCR, translation, file conversion, SMS, email, phone calls, and more — paying per request with Bitcoin Lightning micropayments. No API keys, no accounts, no KYC.
+
+## Preconditions
+
+- A Lightning wallet with sats (any wallet: Phoenix, Wallet of Satoshi, Alby, etc.)
+- Goose with the Sats4AI MCP server configured, OR direct access to the MCP endpoint
+
+### Add to Goose
+
+The Sats4AI server is a remote streamable-http MCP — no install required:
+
+```
+Type: streamable-http
+URL: https://sats4ai.com/api/mcp
+No environment variables needed
+```
+
+Or install via npm:
+
+```bash
+npx sats4ai-mcp
+```
+
+## Payment Flow
+
+Every tool requires a Lightning micropayment before it runs. The flow is always the same:
+
+1. Call `create_payment` with the tool name and any required parameters → get a Lightning invoice + `paymentId`
+2. Pay the invoice with any Lightning wallet (confirms in seconds)
+3. Call `check_payment_status({ paymentId })` → wait for `{ paid: true }`
+4. Call the tool with `{ paymentId, ...params }` → get the result
+
+Payments are single-use. If the service fails, the payment is marked refundable — call `request_refund({ paymentId })`.
+
+## Available Tools
+
+### Discovery
+- `list_models` — browse all tools with pricing (free)
+- `get_model_pricing` — check price for a specific tool (free)
+- `list_planned_services` — see what's coming next (free)
+- `vote_for_service` — vote for a tool to be built (1 sat)
+
+### Image Generation & Editing
+- `generate_image` — text-to-image (Grok Imagine, Seedream 4, Nano Banana 2)
+- `edit_image` — edit images with natural language instructions
+
+### Image Processing
+- `remove_background` — BiRefNet SOTA, handles hair/glass/transparency (5 sats)
+- `upscale_image` — 2x/4x Real-ESRGAN super-resolution (5 sats)
+- `restore_face` — CodeFormer NeurIPS 2022 face restoration (5 sats)
+- `detect_nsfw` — safe/suggestive/explicit classification (2 sats)
+- `detect_objects` — bounding boxes + confidence scores via Grounding DINO (5 sats)
+- `remove_object` — describe what to remove, no mask needed (15 sats)
+- `colorize_image` — black-and-white photo colorization via DDColor (5 sats)
+- `deblur_image` — NAFNet ECCV 2022 blur/noise removal (20 sats)
+
+### Video & 3D
+- `generate_video` — text-to-video with Kling v3, async (per second)
+- `animate_image` — still image to video via Grok Imagine Video, async
+- `generate_3d_model` — photo to textured GLB model via Seed3D, async
+
+### Audio
+- `text_to_speech` — 467 voices, 45+ languages, emotion control, voice cloning
+- `clone_voice` — clone any voice from a single audio sample
+- `transcribe_audio` — Mistral Transcription with speaker diarization + timestamps
+- `generate_music` — AI vocals and instrumentation via Music-1.5
+
+### Documents & Files
+- `extract_document` — PDF/image to Markdown via Mistral OCR
+- `extract_receipt` — structured data from receipts and invoices
+- `convert_file` — 200+ format conversions
+- `merge_pdfs` — merge multiple PDFs preserving bookmarks
+- `convert_html_to_pdf` — HTML or Markdown to pixel-perfect PDF
+- `epub_to_audiobook` — EPUB/PDF/TXT to M4B audiobook with chapters, async
+
+### Text & Translation
+- `generate_text` — frontier LLMs (Kimi K2.5, GPT-OSS-120B, Qwen3-32B)
+- `analyze_image` — image understanding via Qwen VL
+- `translate_text` — 119 languages via Qwen3-32B
+
+### Communication
+- `send_email` — send email to anyone, no SMTP setup
+- `send_sms` — send SMS to any phone number, no telecom account
+- `place_call` — deliver a spoken message to any phone
+- `ai_call` — send an AI voice agent to handle a two-way phone conversation, returns transcript + analysis, async
+- `respond_to_ai_call` — answer clarification questions before an AI call proceeds
+
+### Async Job Lifecycle
+Some tools (video, 3D, transcription, audiobook, AI calls) are async. They return a `jobId` immediately:
+- `check_job_status` — poll until `status: "completed"`
+- `get_job_result` — retrieve the final output URL
+
+### Support
+- `check_payment_status` — confirm a payment was received
+- `request_refund` — request refund for a failed service (free)
+
+## Example Workflow: Remove background from an image
+
+```
+1. create_payment({ toolName: "remove_background" })
+   → { paymentId: "pay_abc", invoice: "lnbc50n1p...", amountSats: 5 }
+
+2. Pay the Lightning invoice with any wallet
+
+3. check_payment_status({ paymentId: "pay_abc" })
+   → { paid: true }
+
+4. remove_background({ paymentId: "pay_abc", imageUrl: "https://example.com/photo.jpg" })
+   → { resultUrl: "https://sats4ai.com/results/photo-nobg.png" }
+```
+
+## Verification
+
+- [ ] `list_models` returns the tool catalog without payment
+- [ ] `create_payment` returns a valid BOLT-11 Lightning invoice
+- [ ] After paying, `check_payment_status` returns `{ paid: true }`
+- [ ] Tool call with the `paymentId` returns the expected result
+
+## Resources
+
+- Docs: https://sats4ai.com/mcp
+- Discovery manifest: https://sats4ai.com/.well-known/mcp
+- Tool catalog with pricing: https://sats4ai.com/api/mcp/discovery
+- npm: https://www.npmjs.com/package/sats4ai-mcp