Menü
Diagnose what is missing before a Trellis Cloudflare app can boot, deploy, or verify live.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Basierend auf der SOC-Berufsklassifikation
Build or modify a Trellis agent by editing the agent blueprint, skills, knowledge, state map, MCP surfaces, and verification commands.
Connect Claude Code, Codex, Cursor, or OpenCode to a local or deployed Trellis Cloudflare app over MCP.
Connect Trellis providers as Cloudflare Worker secrets one provider at a time.
Scaffold a new Cloudflare-first Trellis app and hand the user to readiness or deploy.
Take a first-time Trellis user from local proof to the simplest credible Cloudflare production deployment.
Route a new Trellis user through the Cloudflare-first path without making them understand the whole stack first.
| name | trellis-readiness-check |
| description | Diagnose what is missing before a Trellis Cloudflare app can boot, deploy, or verify live. |
Use this skill when the user wants to know what is missing before Trellis can run.
Prefer JSON:
npm run doctor -- --json
npm run verify -- --json
If the user is not in a generated app but has the CLI script:
npm run trellis -- doctor --json
npm run trellis -- verify cloudflare --json
Group blockers into:
Do not dump the full environment matrix unless the user asks for it.
The fastest proof is:
nvm use 22
npm install
npm run cf:login
npm run doctor -- --json
npm run smoke -- --json
npm run verify -- --json
Success means:
wrangler.jsonc has the expected bindingsThat does not prove:
After deploy:
npm run verify -- --live --url "$APP_URL" --api-key "$TRELLIS_API_KEY"
Only run a live agent exercise when the user accepts that it may use one model call:
npm run verify -- --live --url "$APP_URL" --api-key "$TRELLIS_API_KEY" --exercise-agent
For Attio provider smoke:
npm run verify -- --live --url "$APP_URL" --attio-smoke --provider-smoke-token "$TRELLIS_PROVIDER_SMOKE_TOKEN"
Use precise next-step language:
Blocked: run npm run cf:login or set CLOUDFLARE_ACCOUNT_ID and CLOUDFLARE_API_TOKENBlocked: TRELLIS_DB binding is missing from wrangler.jsoncBlocked: TRELLIS_PACKS bucket_name is missing from wrangler.jsoncBlocked: PROSPECT_WORKFLOW binding is missing from wrangler.jsoncBlocked: set TRELLIS_API_KEY as a Worker secret before protected remote checksNext: run npm run deploy -- --json so Trellis can provision D1, R2, Queues, Workflows, and sync packsNext: use trellis-setup-database to inspect D1 schema, trace rows, approvals, and seeded demo dataLater: connect FIRECRAWL_API_KEY after the Cloudflare app is aliveOnly tell the user the next missing step. Do not switch them into provider setup until doctor, smoke, deploy, and Cloudflare verification are green enough to trust.
When the readiness problem is about app behavior rather than infrastructure, route to trellis-build-agent and inspect:
src/agent.ts imports trellis and schema from @trellis/gtm.src/agent.ts keeps trellis.safeOutbound() unless the user explicitly changed safety.mcp.surface is role-shaped when appropriate, such as sdr.mcp.operator.name is present when an operator MCP should exist.mcp.tools.skillTools maps bounded tool names to real skill directory names.src/state/*.map.ts uses trellis.state(...) instead of hand-authored D1 migrations.skills/**/SKILL.md has purpose, inputs, method, output contract, and safety sections.