Skip to main content

ankra ai

Configure the organisation’s AI provider, credentials, custom OpenAI-compatible endpoints, and the model catalog the chat picker offers. Mutations require the ai.manage permission (organisation admins by default); the provider status and the model listing are available to every member.

ankra ai anthropic

Manage the custom Anthropic API key

ankra ai anthropic delete

Remove the custom Anthropic API key
Flags

ankra ai anthropic set

Store and validate a custom Anthropic API key (must start with sk-ant-).
Flags

ankra ai autonomy

Read and operate the organisation-wide AI kill switch. Engaging it stops all AI for everyone in the organisation: running sessions and agent runs are cancelled, and chat is refused until it is released. Reach for it during an incident, not to tighten a policy - a narrower stop lives in the auto-remediation policy.

ankra ai autonomy pause

Pause autonomous actions across the organisation. Chat, diagnosis and read-only investigation keep working; what stops is the AI changing anything without a person approving it first. It switches off auto-remediation and pauses every enabled agent, and records exactly what it switched off - so resuming restores that and nothing else, from any terminal or browser, by any administrator. To stop AI entirely during an incident, use “ankra ai autonomy stop-all”.
Flags

ankra ai autonomy resume

Resume autonomous actions. Only what the pause switched off is restored: an agent somebody disabled deliberately stays disabled.
Flags

ankra ai autonomy start-all

Let AI run again for the organisation
Flags

ankra ai autonomy status

Show whether AI is stopped for the organisation
Flags

ankra ai autonomy stop-all

Stop all AI for everyone in the organisation. Every running AI session and agent run is cancelled, pending approvals expire, and chat is refused until the switch is released. The reason is shown to administrators alongside the switch.
Flags

ankra ai board-identity

Manage the organisation’s board agent identity. Agents the platform created - the preloaded team - have no user account of their own, so their headless runs cannot change anything. The board identity gives them one, with its own role, so their work is attributable to the agent rather than to whoever set it up. It cannot sign in and has no token. Without it, a designated board worker is escalated to a human on every ticket instead of working it.

ankra ai board-identity provision

Give the board an identity its agents act as
Flags

ankra ai board-identity revoke

Stand the board identity down. The principal itself is kept so its past actions stay attributable; what it loses is every grant above the viewer floor, which makes the board’s agents ineligible again on the next dispatcher pass.
Flags

ankra ai board-identity status

Show whether the board has an identity
Flags

ankra ai endpoints

Manage custom OpenAI-compatible endpoints

ankra ai endpoints create

Add a named OpenAI-API-standard endpoint (OpenRouter, vLLM, LiteLLM, Ollama, …). The server validates and probes the endpoint before saving; the API key is stored server-side and never returned.
Flags

ankra ai endpoints delete

Delete an endpoint and its stored key. Catalog models on it are removed too.
Flags

ankra ai endpoints discover

Query the endpoint’s /models list so you know which model ids to reference.
Flags

ankra ai endpoints list

List configured OpenAI-compatible endpoints
Flags

ankra ai endpoints update

Update an endpoint’s name, base URL, and optionally rotate its key. Omit —api-key to keep the stored key.
Flags

ankra ai models

Manage the organisation model catalog

ankra ai models create

Add a model to the catalog. For an Ankra-managed Claude model pass —model-id (e.g. claude-opus-5). For a model served by a custom OpenAI-compatible endpoint, also pass —endpoint with the endpoint id (see ‘ankra ai endpoints list’).
Flags

ankra ai models delete

Delete a catalog model
Flags

ankra ai models list

List the catalog models the chat picker offers
Flags

ankra ai models reset

Revert the catalog to the built-in defaults
Flags

ankra ai models update

Update a catalog model addressed by its row id or catalog key. Only the flags you pass are changed; the rest keep their current values. Editing a built-in default the first time materialises the catalog.
Flags

ankra ai openai

Manage the legacy single OpenAI-compatible endpoint configuration. For multiple endpoints and per-model routing prefer ‘ankra ai endpoints’ and ‘ankra ai models’; this command configures the single legacy endpoint the ‘openai_compatible’ provider uses.

ankra ai openai delete

Remove the legacy OpenAI-compatible endpoint
Flags

ankra ai openai set

Store the legacy OpenAI-compatible endpoint
Flags

ankra ai openrouter

Manage the organisation’s OpenRouter API key (bring your own key). Store a key with ‘set’, then activate it with ‘ankra ai provider openrouter’. Removing the key while OpenRouter is the active provider resets the organisation to the Ankra-managed default.

ankra ai openrouter remove

Remove the organisation’s OpenRouter API key. When OpenRouter is the active provider the organisation falls back to the Ankra-managed default.
Flags

ankra ai openrouter set

Store and validate an OpenRouter API key (must start with sk-or-). Pass the key with —api-key, pipe it on stdin, or omit both to be prompted interactively. The interactive prompt masks the input; the key is never echoed or logged.
Flags

ankra ai provider

Set the active AI provider for the organisation. ankra Ankra-managed Claude models (the default). anthropic Your own Anthropic API key (set it first with ‘ankra ai anthropic set’). openai_compatible Your OpenAI-compatible endpoint (set it first with ‘ankra ai openai set’). openrouter Your own OpenRouter API key (set it first with ‘ankra ai openrouter set’).
Flags

ankra ai status

Show the active AI provider and configured credentials
Flags