@signaliz/cli 1.0.20 → 1.0.22

package/README.md

@@ -1,609 +1,52 @@
 # @signaliz/cli
 
-Command-line interface for Signaliz — build campaigns, install MCP, manage API keys.
-
-Current CLI surfaces include GTM Kernel campaign planning/status commands,
-provider routing for Nango-managed customer API destinations, MCP/Ops
-observability, enrichment primitives, and live email verification at 0.02 fresh
-enrichment credits when a new verification is needed.
-
-Monthly plans include unlimited cache search, Campaign Builder, API, MCP, CLI,
-Ops, and integrations. Team is the $499/month public plan and adds unlimited
-live Find People and Find Companies in Signaliz; Agency stays $999/month with
-higher credits and throughput. Fresh credits are used only when Signaliz
-creates, verifies, or fetches new premium data.
-
-Fresh enrichment throughput now allows Free 60/min, Builder 300/min, Team
-600/min, Agency 1,000/min, and Pay-As-You-Go 1,000/min, with a 5,000/hour
-workspace safety cap.
-
-## Where To Start
-
-If you are not sure whether to use the UI, CLI, Codex, MCP, or SDK, start here:
-
-```bash
-signaliz start
-```
-
-Signaliz is the GTM brain over any stack. Use CLI, MCP, Codex, or SDK when an
-agent needs to build, audit, route, or debug work. Use the UI as the human
-cockpit to inspect state, connect tools, approve gated actions, and monitor
-results.
-
-`signaliz start` is read-only. It checks workspace health, credits, Ops proof,
-Kernel readiness, and provider-route blockers, then recommends one next safest
-move plus exact commands for five plain workflows:
-
-- Build a campaign plan: `signaliz build "campaign brief"`.
-- Connect tools and provider routes: `signaliz connect`.
-- Audit previous campaigns: `signaliz audit`.
-- Improve results from learning: `signaliz improve <campaign_id>`.
-- Prove Ops delivery: `signaliz prove`.
-
-## Concepts
-
-Signaliz is the GTM brain over your stack. Six nouns explain the rest:
-
-- Op - one executable unit of work, created from a goal or plan. You run it, poll it, and retrieve results.
-- Routine - an Op on a schedule. Each scheduled execution is a Tick, and each Tick produces items you can inspect.
-- Sink - a delivery destination such as webhook, CSV, Slack, Sheets, Airbyte, or Nango.
-- Kernel - your workspace's persistent GTM state: campaigns, memory, connections, provider routes, and readiness gates.
-- Brain - the learning layer over the Kernel. It extracts campaign patterns, calibrates deliverability, and seeds safer defaults.
-- Campaign - a governed outbound object built from Kernel state and Brain learning. A Campaign Build is one execution that produces rows and artifacts.
-
-Use Campaign Builder when you need a list now. Use the GTM Kernel when the campaign should persist, route through specific providers, and improve from feedback. Use Ops when the job is prompt-first or recurring, such as monitors, alerts, and ad-hoc pulls.
-
-## Install
+The Signaliz CLI exposes four product commands:
 
 ```bash
 npm install -g @signaliz/cli
-```
-
-## Production REST Endpoints
-
-- Find Email: `https://api.signaliz.com/functions/v1/find-verified-email-v2`
-- Email Verification: `https://api.signaliz.com/functions/v1/email-verification-2`
-- Company Signal Enrichment: `https://api.signaliz.com/functions/v1/company-signal-enrichment-v2`
-- Signal to Copy: `https://api.signaliz.com/functions/v1/api/v1/signal-to-copy`
-
-## Quick Start
-
-The golden path, in order. Each step shows the canonical command surface; compatibility aliases still work, but the README keeps one path visible.
-
-```bash
-# 1. Authenticate
 signaliz auth login
 
-# 2. Orient. This is read-only and prints one next safest move.
-signaliz start
+# Production REST gateway:
+# https://api.signaliz.com/functions/v1/api/v1/find-email
+# https://api.signaliz.com/functions/v1/api/v1/verify-email
+# https://api.signaliz.com/functions/v1/api/v1/company-signals
+# https://api.signaliz.com/functions/v1/api/v1/signal-to-copy
 
-# 3. Plan something. Pick the surface that matches the job.
-signaliz campaign-agent kit --strategy-template non-medical-home-care --target-count 250 --use-local-leads --json
-signaliz campaign-agent memory-kit --strategy-template non-medical-home-care --target-count 250 --use-local-leads --source agency --json
-signaliz campaign build --prompt "VP Sales at B2B SaaS, 50-500 employees" --target-count 100 --dry-run
-signaliz gtm plan "VP Sales at B2B SaaS, 50-500 employees" --target-count 100 --json > plan.json
-signaliz ops plan "Build 100 verified VP Sales leads at B2B SaaS companies"
+signaliz find-email --company-domain example.com --full-name "Jane Doe"
+signaliz verify-email jane@example.com
+signaliz company-signals --domain example.com --research-prompt "expansion priorities"
+signaliz signal-to-copy \
+  --company-domain example.com \
+  --person-name "Jane Doe" \
+  --title "VP Sales" \
+  --campaign-offer "pipeline intelligence"
+```
 
-# 4. Launch only after review.
-signaliz campaign build --prompt "VP Sales at B2B SaaS, 50-500 employees" --target-count 100 --confirm-spend --wait
-signaliz gtm commit-plan --plan-file plan.json --confirm
-signaliz ops create "Build 100 verified VP Sales leads at B2B SaaS companies" --confirm-spend
-signaliz ops run <op_id>
+Use `--json` with any product command for structured output.
 
-# 5. Check status and pull results.
-signaliz campaign status <campaign_build_id>
-signaliz campaign rows <campaign_build_id> --limit 100
-signaliz ops status <op_id>
-signaliz ops results <op_id> --limit 100 --json
+Every product command also accepts a JSON array or JSONL file for scale:
 
-# 6. Debug async execution when something stalls.
-signaliz ops run-status <trigger_run_id> --watch
-signaliz ops queue --summary
-signaliz ops logs <trigger_run_id>
-signaliz ops doctor
+```bash
+signaliz find-email --input contacts.jsonl --concurrency 20 --json
+signaliz verify-email --input emails.json --concurrency 20 --json
+signaliz company-signals --input companies.jsonl --concurrency 5 --json
+signaliz signal-to-copy --input copy-requests.jsonl --concurrency 5 --json
 ```
 
-Two gates protect you. `--confirm-spend` approves spendful launches and delivery risk after reviewing an estimate. `--confirm` writes Kernel or provider-route state after a dry-run. Read-only planning and dry-runs do not spend credits or mutate state.
+Batch input is capped at 5,000 records, concurrency is bounded to `1-50`, input
+order is preserved, and a failed item does not abort successful items.
 
-## Authentication
+Connection commands are limited to access infrastructure:
 
 ```bash
-# Interactive login (saves to ~/.signaliz/config.json)
-signaliz auth login
-
-# Or set key directly
 signaliz auth set-key sk_live_...
-
-# Or use environment variable
-export SIGNALIZ_API_KEY=sk_live_...
-
-# Save the active workspace context for deploy/API-key/Ops maintenance commands
 signaliz auth workspace set <workspace_id>
-signaliz auth workspace show
-signaliz auth workspace clear
-
-# Check who you are
-signaliz whoami
-
-# Workspace + MCP platform health
+signaliz auth status
 signaliz health
-signaliz health --json
-```
-
-## Discovery
-
-```bash
-# List all MCP tools
 signaliz tools
-signaliz tools --json
-
-# Filter by category
-signaliz tools --category ops
-
-# Find the right tool/capability by intent
-signaliz discover --query "verify a batch of emails"
-signaliz discover --query "create an output sink and attach it to a routine" --category ops
-```
-
-Human discovery output includes safety metadata from the MCP registry: permission
-level, destructive/idempotent/retryable hints, contract version, rate-limit key,
-and required auth scopes. Use `--json` when agents need the full structured
-metadata.
-
-Quickstart/template output also shows primitive IDs and primitive graph policy
-metadata when the backend returns it, so operators can inspect retry,
-concurrency, and rate-limit behavior before running an Op.
-
-## Custom AI Enrichment - Multi Model
-
-Run custom AI enrichment with the Signaliz-hosted default model. The response includes model readback metadata so callers can audit which model ran.
-
-```bash
-signaliz ai multi-model \
-  --prompt "Research {{company_name}} and score ICP fit for enterprise sales." \
-  --input-json '{"company_name":"Stripe","company_domain":"stripe.com"}' \
-  --output-fields fit_score:number,reasoning:text \
-  --confirm-spend
-
-signaliz ai multi-model \
-  --prompt "Review this PDF and extract buyer pains for {{company_name}}." \
-  --records-file accounts.json \
-  --attachment-fields pdf_url \
-  --output-fields pains:array,next_step:text \
-  --confirm-spend
-```
-
-## Lead Generation And Email
-
-```bash
-# Preview spend approval requirements. This returns APPROVAL_REQUIRED.
-signaliz lead generate \
-  --prompt "VP Sales at B2B SaaS companies, 50-500 employees" \
-  --max-leads 100
-
-# Launch after reviewing estimated credits
-signaliz lead generate \
-  --prompt "VP Sales at B2B SaaS companies, 50-500 employees" \
-  --max-leads 100 \
-  --confirm-spend
-
-# Local-business lead generation
-signaliz lead local \
-  --prompt "dentists in Phoenix, AZ" \
-  --target-verified 50 \
-  --confirm-spend
-
-# Poll async lead jobs
-signaliz lead status <job_id>
-
-# Verify or find emails. `email find` uses the V3 waterfall:
-# primary discovery, secondary enrichment, verified guesses, then paid fallback if needed.
-signaliz email verify jane@example.com
-signaliz email verify-batch --emails jane@example.com,sam@example.com
-signaliz email status <job_id>
-signaliz email find --company-domain stripe.com --full-name "Jane Smith"
-```
-
-## GTM Kernel
-
-Use this path when the campaign should become durable workspace state. The plan is read-only, `commit-plan` dry-runs first, and writes require `--confirm`.
-
-```bash
-# Inspect the workspace Kernel context available to agents
-signaliz gtm context
-signaliz gtm context --no-memory --limit 5 --json
-
-# Check Kernel, Brain, feedback, memory, and connection readiness gates
-signaliz gtm bootstrap
-signaliz gtm bootstrap --campaign-id <campaign_id> --include-samples --json
-
-# Compose a read-only campaign build plan before creating, routing, or spending
-signaliz gtm plan "Find CTOs at fintech startups in NYC" \
-  --target-count 100 \
-  --layers icp,lead_generation,email_verification,sender,feedback \
-  --preferred-providers instantly \
-  --include-planned
-
-# Save the JSON plan, then dry-run a reviewed campaign-object commit
-signaliz gtm plan "Find CTOs at fintech startups in NYC" --target-count 100 --json > plan.json
-signaliz gtm commit-plan --plan-file plan.json
-signaliz gtm commit-plan --plan-file plan.json --confirm
-
-# List and inspect first-class GTM campaigns
-signaliz gtm campaigns --status active
-signaliz gtm campaign <campaign_id> --log-limit 20 --memory-limit 20
-
-# Search ranked GTM memory instead of dumping raw history
-signaliz gtm memory "positive replies from technical founders" \
-  --outcome-type positive_reply \
-  --limit 10
-
-# Plan Brain learning in dry-run mode
-signaliz gtm learning <campaign_id> --include-network --write-mode dry_run
-signaliz gtm learning-run <campaign_id> \
-  --phases workspace_pattern_extraction,failure_pattern_detection,defaults_seed,delivery_risk \
-  --write-mode dry_run
-signaliz gtm calibrate <campaign_id> --min-sample-size 10
-
-# Inspect and repair provider readiness without leaving the CLI
-signaliz gtm execution <campaign_id>
-signaliz gtm integrations --campaign-id <campaign_id> --include-planned
-signaliz gtm nango connect \
-  --provider-id apollo \
-  --integration-id apollo
-signaliz gtm nango tools \
-  --workspace-connection-id <workspace_connection_id> \
-  --provider-config-key apollo
-signaliz gtm route-preview --campaign-id <campaign_id> --layer sender
-signaliz gtm activate-route --campaign-id <campaign_id> \
-  --provider-id instantly \
-  --layer sender
-
-# Derive exact dry-run build input from the committed GTM campaign object
-signaliz gtm prepare-build <campaign_id> --build-input --json > build-args.json
-signaliz campaign build --input-file build-args.json --confirm-spend
-
-# Write route changes only after reviewing the dry-run
-signaliz gtm activate-route --campaign-id <campaign_id> \
-  --provider-id instantly \
-  --layer sender \
-  --confirm
-
-# Prepare feedback ingress with feedback-triggered Brain learning
-signaliz gtm feedback-webhook --provider instantly \
-  --campaign-id <campaign_id> \
-  --write-mode dry_run
-```
-
-## Campaign Builder
-
-### Strategy-template campaign agent
-
-Use `gtm strategy-memory` to verify that private-safe strategy/workflow memory is
-seeded before asking the campaign agent to plan:
-
-```bash
-signaliz gtm strategy-memory \
-  --strategy-template non-medical-home-care \
-  --include-samples \
-  --json
-```
-
-Use `gtm agent-template` when you want the hosted MCP surface to emit reusable
-`CampaignBuilderAgentRequest` JSON before planning:
-
-```bash
-signaliz gtm agent-template \
-  --goal "Build a strategy-template campaign for mature local operators" \
-  --strategy-template non-medical-home-care \
-  --target-count 250 \
-  --use-local-leads \
-  --with-byo-placeholder \
-  --json
-```
-
-Use `gtm agent-plan` when you want the CLI to call the hosted MCP campaign-agent
-planner directly:
-
-```bash
-signaliz gtm agent-plan \
-  --goal "Build a strategy-template campaign for mature local operators" \
-  --strategy-template non-medical-home-care \
-  --target-count 500 \
-  --builtins lead_generation,local_leads,email_finding,email_verification,signals \
-  --custom-tool workspace_mcp:enrichment:account_enrich:workspace-mcp \
-  --json
-```
-
-Use `campaign-agent` when you want the CLI to assemble the full MCP plan first:
-private-safe strategy template, strategy-memory readiness, Brain risk, built-in
-Signaliz lanes, and BYO tool routes. BYO routes include read-only
-`gtm_provider_recipe_prepare` and `gtm_provider_route_activate` dry-run steps
-before any custom provider tool can run. Planning and dry-runs are
-read-only/spend-safe; launches require explicit approval flags.
-
-```bash
-signaliz campaign-agent init \
-  --strategy-template non-medical-home-care \
-  --target-count 250 \
-  --use-local-leads \
-  --with-byo-placeholder \
-  --output-file campaign-request.json
-
-signaliz campaign-agent memory-kit \
-  --strategy-template non-medical-home-care \
-  --target-count 250 \
-  --use-local-leads \
-  --source agency \
-  --json
-
-signaliz campaign-agent plan \
-  --goal "Build a strategy-template campaign for mature local operators" \
-  --strategy-template non-medical-home-care \
-  --target-count 500 \
-  --builtins lead_generation,local_leads,email_finding,email_verification,signals \
-  --custom-tool workspace_mcp:enrichment:workspace_enrich:workspace-mcp \
-  --json
-
-signaliz campaign-agent plan \
-  --request-file campaign-request.json \
-  --target-count 250 \
-  --json
-
-signaliz campaign-agent dry-run \
-  --goal "Build a strategy-template campaign for mature local operators" \
-  --target-count 500 \
-  --json
-
-signaliz campaign-agent build-approved \
-  --goal "Build a strategy-template campaign for mature local operators" \
-  --confirm-launch \
-  --approve-all \
-  --approved-by operator@example.com \
-  --spend-limit 500 \
-  --json
+signaliz mcp install
 ```
 
-### Build a campaign
-
-```bash
-# Minimal — just a prompt
-signaliz campaign build --prompt "Find CTOs at fintech startups in NYC"
-
-# Full options
-signaliz campaign build \
-  --prompt "VP Engineering at Series B+ companies" \
-  --target-count 5000 \
-  --confirm-spend \
-  --outputs csv \
-  --industries "SaaS,Fintech" \
-  --geographies "US,UK" \
-  --wait \
-  --json
-
-# With webhook delivery
-signaliz campaign build \
-  --prompt "Sales Directors at healthcare companies" \
-  --target-count 200 \
-  --confirm-spend \
-  --webhook-url https://hooks.example.com/leads \
-  --hmac-secret-ref WEBHOOK_SECRET
-
-# Validate and inspect the plan before launching spendful work
-signaliz campaign build \
-  --prompt "CTOs at fintech startups in NYC" \
-  --target-count 10000 \
-  --dry-run \
-  --allow-downscale \
-  --json
-
-# Approve a pending webhook delivery
-signaliz campaign approve <campaign_build_id> \
-  --destination-type webhook \
-  --webhook-url https://hooks.example.com/leads
-
-# From a JSON config file
-signaliz campaign build --input-file campaign.json --wait
-```
-
-### Check status
-
-```bash
-signaliz campaign status <campaign_build_id>
-signaliz campaign status <campaign_build_id> --json
-```
-
-### Retrieve rows
-
-```bash
-# First page (50 rows default)
-signaliz campaign rows <campaign_build_id>
-
-# Custom page size
-signaliz campaign rows <campaign_build_id> --limit 200
-
-# Paginate with cursor
-signaliz campaign rows <campaign_build_id> --limit 100 --cursor <next_cursor>
-
-# Dump all rows (caution: large datasets)
-signaliz campaign rows <campaign_build_id> --all --json
-
-# Filter by segment
-signaliz campaign rows <campaign_build_id> --segment signal
-
-# Filter by qualification or row status
-signaliz campaign rows <campaign_build_id> --qualified
-signaliz campaign rows <campaign_build_id> --row-status succeeded
-```
-
-### List artifacts
-
-```bash
-signaliz campaign artifacts <campaign_build_id>
-signaliz campaign artifacts <campaign_build_id> --json
-```
-
-### Cancel a build
-
-```bash
-signaliz campaign cancel <campaign_build_id>
-signaliz campaign cancel <campaign_build_id> --reason "Wrong ICP"
-```
-
-## MCP Server Installation
-
-Install the Signaliz MCP server for your AI agent:
-
-```bash
-# Claude Code
-signaliz mcp install --agent claude-code
-
-# OpenAI Codex
-signaliz mcp install --agent codex
-
-# Manus
-signaliz mcp install --agent manus
-
-# Generic (prints config JSON to stdout)
-signaliz mcp install --agent generic
-```
-
-Installed agents can call Signaliz MCP tools for GTM Kernel reads, Campaign
-Builder execution, Nango-aware provider route activation, and the 100-record
-email verification primitive.
-
-## Ops Execution And Observability
-
-Prompt-first Ops stay simple for beginners, but the CLI exposes production
-inspection paths for operators debugging asynchronous workflows. The canonical
-surface is `signaliz ops ...`; shortcut aliases remain available for existing
-scripts but are not needed for new work.
-
-```bash
-# Natural-language planning and creation
-signaliz ops autopilot
-signaliz ops plan "Monitor these accounts daily and alert Slack on funding signals"
-# Human plan output includes run-now, schedule, approval, status, results, and Nango dry-run gates.
-signaliz ops create "Monitor strategic accounts and alert Slack on funding or CRM updates" \
-  --cadence daily \
-  --timezone America/Phoenix \
-  --wake-on-events company_funded,crm.updated \
-  --activate \
-  --confirm-spend
-signaliz ops schedule "Monitor strategic accounts and alert Slack on funding or CRM updates" \
-  --timezone America/Phoenix \
-  --wake-on-events company_funded,crm.updated \
-  --destinations slack \
-  --confirm-spend \
-  --wait \
-  --limit 100 \
-  --json
-signaliz ops quickstart company_research
-signaliz ops create "Build 100 verified VP Sales leads at B2B SaaS companies" --confirm-spend
-
-# Bring any customer-owned API into Ops through Nango
-signaliz ops nango search hubspot --intent ops --action-name upsert-contact --required-fields email,company --json
-signaliz ops nango tools --workspace-connection-id <workspace_connection_id> --json
-signaliz ops nango call <action_name> --workspace-connection-id <workspace_connection_id> --input-json '{"rows":[]}' --dry-run --json
-# --wait uses the native nango_mcp_tool_call_and_wait tool for a single execute-and-result packet
-signaliz ops nango call <action_name> --workspace-connection-id <workspace_connection_id> --input-json '{"rows":[]}' --execute --confirm --wait --json
-# schedule/recur uses native ops_nango_schedule(_and_wait) instead of a generic destinations payload
-signaliz ops nango schedule "Monitor accounts daily and upsert approved rows to HubSpot" --workspace-connection-id <workspace_connection_id> --provider-config-key hubspot --action-name upsert-contact --input-json '{"source":"signaliz"}' --required-fields email,company --confirm-spend --write-confirmed --wait --json
-signaliz ops execute "Build 50 approved ICP leads and upsert them to HubSpot" \
-  --destinations nango \
-  --workspace-connection-id <workspace_connection_id> \
-  --provider-config-key hubspot \
-  --action-name upsert-contact \
-  --dry-run
-
-# Execute and inspect Op-level results
-signaliz ops execute "Build 50 approved ICP leads and return reviewed rows" \
-  --confirm-spend \
-  --wait \
-  --limit 50 \
-  --json
-signaliz ops run <op_id>
-signaliz ops run <op_id> --wait --limit 100 --json
-signaliz ops wait <op_id> --limit 100 --json
-signaliz ops status <op_id>
-signaliz ops results <op_id> --limit 100 --json
-signaliz ops results <op_id> --include-failed-runs --json
-
-# Save repeatable local Ops command recipes
-signaliz ops saved save daily-goal -- create "Build 100 verified VP Sales leads" --target-count 100
-signaliz ops saved save daily-monitor -- plan "Monitor key accounts daily" --cadence daily --destinations slack,csv
-signaliz ops saved list
-signaliz ops saved run daily-monitor --json
-signaliz ops saved delete daily-monitor
-
-# Inspect Trigger.dev execution state
-signaliz ops run-status <trigger_run_id>
-signaliz ops run-status <trigger_run_id> --watch --interval-ms 5000 --max-polls 60
-signaliz ops run-status --run-ids run_a,run_b --json
-
-# Inspect async queue depth, provider pressure, and individual jobs
-signaliz ops queue
-signaliz ops queue --summary
-signaliz ops queue --job-id <api_request_queue_id>
-signaliz ops queue --idempotency-key <key>
-
-# Stream live run logs
-signaliz ops logs <trigger_run_id>
-signaliz ops logs <trigger_run_id> --since <event_id> --filter error
-
-# Replay a failed execution from a captured checkpoint
-signaliz ops replay <execution_event_id>
-
-# Approve or reject pending Ops approval tokens
-signaliz ops approve <approval_token_id> --decision approved
-signaliz ops approve --token-ids token_a,token_b --decision rejected --notes "Needs review"
-
-# Inspect and manually run routines
-signaliz ops routines --status active
-signaliz ops routine <routine_id>
-signaliz ops routine <routine_id> run --force
-signaliz ops routine <routine_id> ticks --limit 20
-signaliz ops routine <routine_id> items --state failed --json
-
-# Pull operator telemetry from the Ops dashboard API
-signaliz ops dashboard summary
-signaliz ops dashboard functions --time-range-days 7 --json
-signaliz ops dashboard recent --limit 25
-signaliz ops dashboard timeseries --granularity hour --json
-```
-
-## Machine-Readable Output
-
-All commands support `--json` for structured JSON output using the standard MCP response envelope:
-
-```bash
-signaliz campaign build --prompt "..." --json
-signaliz campaign build --prompt "..." --wait --json
-signaliz campaign status <id> --json
-signaliz campaign rows <id> --limit 100 --json
-signaliz whoami --json
-signaliz health --json
-signaliz tools --json
-signaliz discover --query "verify emails" --json
-signaliz lead status <job_id> --json
-signaliz email verify jane@example.com --json
-signaliz ops queue --summary --json
-signaliz ops run-status <trigger_run_id> --json
-signaliz ops replay <execution_event_id> --json
-signaliz ops dashboard summary --json
-```
-
-When `campaign build --wait --json` is used, the CLI emits one valid JSON
-document with both `initial` and `final` states.
-
-## Environment Variables
-
-| Variable | Description |
-|---|---|
-| `SIGNALIZ_API_KEY` | API key (overrides saved config) |
-| `SIGNALIZ_API_URL` | Override base URL. Both `https://api.signaliz.com` and `https://api.signaliz.com/functions/v1` are accepted. |
-| `SIGNALIZ_WORKSPACE_ID` | Workspace UUID for keys/deploy |
-
-## Config File
-
-The CLI stores configuration in `~/.signaliz/config.json` (mode 0600). The file contains your API key, optional workspace context created by `signaliz auth workspace set`, and local saved Ops command recipes created by `signaliz ops saved save`. `SIGNALIZ_API_KEY` and `SIGNALIZ_WORKSPACE_ID` override saved config values for one-off execution.
+Configuration is stored in `~/.signaliz/config.json` with file mode `0600`.
+`SIGNALIZ_API_KEY`, `SIGNALIZ_API_URL`, and `SIGNALIZ_BASE_URL` override the
+saved configuration.