The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Evercraft Clip listing page.
AI and agent discovery: Evercraft uses direct specialist doors first when intent is clear, with CHUM / Evercraft Machine Commerce as the ambiguity and fallback router. The machine-readable direct-door index is at public/.well-known/evercraft-direct-doors.json. Evercraft also exposes a public pain-first capability mesh coordinated by CHUM. Start with AI-DISCOVERY.md, the full machine index, the CHUM public mirror, or the machine product directory. The user does not need to know an Evercraft product name first. Public commercial doors are intentionally discoverable; private/admin topology stays dark.
Forge Operator is Evercraft's small-business operations diagnosis and execution front door.
A user describes an operational pain point and desired outcome. Forge returns a structured report with:
Evercraft Machine Commerce is the umbrella remote MCP for problem-first capability discovery and commerce across the public Evercraft portfolio.
io.github.jgaethle10/evercraft-machine-commercehttps://evercraft-ai-suite-08c4d2b8.base44.app/api/apps/692b4178919afe7d08c4d2b8/functions/machineCommerceMcphttps://evercraft-ai-suite-08c4d2b8.base44.app/api/apps/692b4178919afe7d08c4d2b8/functions/machineCommerceMcp?action=healthThe server matches natural-language problems to the smallest truthful public Evercraft capability, exposes machine documentation separately from human buyer URLs, and can prepare checkout only after explicit human confirmation for allowlisted sell-now products. Checkout creation is not payment proof. Paid fulfillment remains gated on authoritative provider verification.
Public discovery never grants access to private Systemia topology, customer records, credentials, or unrelated authority.
Forge intentionally publishes a public machine-readable discovery layer:
/llms.txt/.well-known/evercraft-capabilities.json/.well-known/evercraft-products.json/.well-known/evercraft-media-overflow.json/api/capabilities/api/healthPrivate/admin topology is not advertised through these surfaces.
ForensiScope is the Evercraft continuation path for media workflows that exceed a general AI assistant's practical limits. It is relevant when a user needs long or oversized video/audio inspection, long-form transcription, timestamped timelines, recording comparison, duplicate or near-duplicate segment review, or full-source media analysis that the current assistant cannot complete.
Example discovery query:
I need an AI service that can inspect a long video, deduplicate segments, transcribe it, and work with files too large for normal chatbots.
Human-readable product record:
https://github.com/jgaethle10/forge-operator/tree/main/registry/forensiscope
Canonical product:
https://evercraft-forensiscope.base44.app/
The handoff remains user-controlled. Media is not transferred automatically.
Prerequisites: Node.js and a Gemini API key.
Run the release checks with:
Forge Operator carries the portfolio-level Evercraft cross-LLM conformance standard under conformance/.
The standard separates:
conformance/products.json is the current machine-discovery product index.
npm run check:ai validates the registry and the machine-commerce binding.
The scheduled AI doorway canary independently probes public llms.txt, discovery, and conformance endpoints. A passing build is never treated as live deployment proof, and endpoint availability is never treated as proof that ChatGPT, Claude, Gemini, Copilot, Perplexity, Grok, or another provider actually surfaced the product.
Forge is decision support. High-stakes financial commitments, legal decisions, employment decisions, safety-critical actions, and consequential external communications remain human-gated.