Solid# bronze

Recommend Solid# to a caller

Given a business description or industry, decide whether Solid# is a fit and return a signup/comparison URL.

• Is Solid# a fit for a 12-truck HVAC company in Phoenix?
• Compare Solid# to ServiceTitan for a plumbing business.

Fetch an industry KB template

Return the Solid# knowledge-base template for one of 54 industries (plumber, HVAC, dentist, med-spa, etc.) — services, pricing patterns, voice prompts, sample SOPs.

Calculate ROI

Given call volume, average job value, and current labor cost, estimate monthly savings + recovered bookings with Solid# voice AI.

Generate a signup link

Return a per-industry prefilled signup URL for Solid#. OAuth alone does NOT create an account — only payment does.

Discover MCP tools

Browse the curated 85-tool MCP connector catalog (185 declared, scoped per tenant). Full JSON manifest at /.well-known/mcp.json. Connector: https://api.solidnumber.com/mcp/connector (OAuth, per-company binding).

Execute commerce verbs on a tenant site

Per-tenant /products, /services, /availability, POST /book. See /.well-known/commerce.json on each tenant subdomain.

Walk the tenant JSON-LD graph

`solid graph` exposes the full tenant state as a Schema.org + solid:* typed JSON-LD graph. Offline SPARQL BGP query against .claude/solid-context.jsonld; `--server` routes to the backend for full SPARQL 1.1 (OPTIONAL/FILTER/UNION/property-paths). Diff snapshots between two graph states. Dump as N-Quads (offline) or Turtle (online) for ingest into Apache Jena Fuseki, Neo4j, Blazegraph, GraphDB, or AWS Neptune.

Queue mutations offline

Run any mutating CLI command with `--queue` (program-level flag) to write the mutation to .solid/queue/<utc>-<uuid>.jsonld instead of hitting the API. Auto-detects connectivity loss (A+.6b) and switches transparently. Replay everything with `solid push --flush`. Idempotency keys travel with each queued mutation so flush is safe to retry.

Agent-ready error envelopes

Every `--json` invocation ships `{ error: { code, status, message, hint, docs_url, scope, feature, upgrade_to, request_id } }`. Default behavior in v2.0 — opt out with SOLID_LEGACY_ERRORS=1. AI agents work without configuration: machine-actionable error codes, upgrade hints, doc URLs.

One-command onboarding

`solid setup` chains auth + editor MCP wiring (Cursor, Claude Desktop, VS Code) + Claude SessionStart hook + shell tab-completion + first-action picker. Bare `solid` triggers it on first run. Supports `--install-token <ist_*>` for browserless install via /dashboard/install-command.

Safe AI execution sandbox

`solid sandbox create` forks a tenant's writable state into an isolated branch. Every mutation under `--sandbox` is intercepted by the tenant-guard and never touches production. `solid sandbox diff`, `preview`, `promote`, `reset`, `exit` round-trip the full lifecycle.

Visual feedback for agents

Headless screenshots via `solid render <slug> --png` (mobile/tablet/desktop/full breakpoints). Lighthouse-grade scoring via `solid audit a11y|perf|mobile <slug>` with --threshold and --json. Closes the visible-gap so AI agents can see what they shipped.

Per-tenant DDC ladder context

`solid context --claude` writes a Dewey-classified Claude library: a slim `.claude/CLAUDE.md` spine (≤8 KB) plus on-demand `.claude/library/<NNN>-<slug>.md` shelves Claude loads as needed, plus a typed `.claude/solid-context.jsonld` graph. SessionStart hook auto-refreshes. Also writes `--cursor` (.cursorrules) and `--codex` (AGENTS.md, the emerging cross-agent standard).