Platform admin

Platform settings

Global engine selection and credentials. Changes apply to every workspace immediately.
🤖 Back End Engine
Pick which AI provider powers content generation across PostAgent. Initial default is Claude (Anthropic) using the existing Config.php token. Switching here updates every workspace at once.
Anthropic Claude — current default. Existing API key in Config.php is used unless you override below.
Model identifier. For Anthropic: claude-sonnet-4-20250514, claude-opus-4-20250514. For OpenAI: gpt-4o, gpt-4-turbo. For Gemini: gemini-1.5-pro.
Provider API key. Encrypted at rest. Leave blank to keep the current value.
🏦 Anthropic billing integration (optional)
Add an Anthropic Admin API key to surface real USD spend and token totals from console.anthropic.com. This is what's actually charged to your Anthropic account, separate from PostAgent's internal counters.
Open Console →
📋 How to generate an Admin API key
  1. Open console.anthropic.com and log in to the account that owns your Anthropic organization.
  2. Click your username at the bottom-left of the page → Organization settings.
  3. In the left navigation, click Admin Keys.
  4. Click + Create admin key → name it (e.g. PostAgent dashboard) → click Add.
  5. Copy the key now — it's shown exactly once. Prefix is sk-ant-admin…. Paste it into the field below and Save.
⚠ The Admin Keys tab is only visible if your Anthropic account is the organization owner. If you don't see it, ask whoever owns the Anthropic org to issue the key (or to promote your account to Admin role under Organization settings → Members). Admin keys grant read-write across the whole org — treat them as sensitive.
Different from the generation key above. Encrypted at rest. Only used to fetch the real-spend numbers shown below — never sent in normal generation calls.
📨 Notification SMTP
Where platform-level emails come from — e.g. credentials for auto-created workspace owners on the New workspace flow. Different from the per-workspace newsletter SMTP.
⚠ Provider switching note
What stays consistent vs what changes when you flip provider.
What works after a switch: all generate / refine / chat / suggest endpoints route to the new provider on the next request. No restart needed.
What needs adjusting: the model string must match the new provider's catalog. Anthropic ↔ OpenAI model names are not interchangeable.
What's NOT switched: historical token_usage_log rows still reflect their original provider (used by the dashboard's "actual consumption" stats). Only new calls hit the new provider.
Custom mode: set the URL of any OpenAI-compatible /v1/chat/completions endpoint. Useful for self-hosted Ollama, Together AI, etc.