Chat Commands
Type / as the first character of a message and the line is a command to the panel rather than a
message to Ora. Type / on its own and the panel lists every command; keep typing and the list narrows
to what you mean. Commands are case-insensitive, and anything after the command name is its argument.
An unrecognised command is not sent to the model — the panel tells you and shows the list.
Conversation
| Command | What it does |
|---|---|
/clear | Reset the conversation and start fresh. |
/reset | Clear all history and working memory. |
/compact | Compact the conversation now: the oldest turns are summarised and dropped. If Ora's last message was not a question, it continues from where it was. |
/export | Export the conversation. /export markdown (the default) gives a readable transcript; /export json gives the raw record. |
/help | List all commands. |
Plan mode
| Command | What it does |
|---|---|
/plan | Put the next request into plan mode: nothing that changes your project runs until you approve a plan. |
/plan <request> | Send that request straight into planning. |
/do | Leave plan mode; the next request runs as an ordinary task. |
/do <request> | Run that request as an ordinary task, without planning. |
See Plan Mode for what approval does and does not allow.
Context and cost
| Command | What it does |
|---|---|
/estimate | Show the current context estimate — how much of the model's window the conversation is using. |
/tokens | Show the same total broken down by category, so you can see what the conversation is mostly made of. |
/context | Show the Studio context snapshot Ora is currently reading: the active view, the open project, suite, case and operation. |
/tools | List the tools Ora can call from here. |
Token figures are estimates produced on this machine, meant for judging scale and trend rather than for billing. Text in different scripts is counted at different rates, so a Chinese conversation reads heavier than an English one of the same length — which is the direction you want when deciding whether there is room left.
Model and style
| Command | What it does |
|---|---|
/model | List the models in the active profile; /model <id> switches to one. |
/thinking | Set the thinking effort: off, low, medium, high, maximum. What a model offers is its own list. |
/personality | Set the interaction style for the conversation. |
/profile | Switch the active LLM profile (your own keys, a managed plan, or an organisation profile). |
Saved conversations
| Command | What it does |
|---|---|
/conversations | List recent conversations. Each row shows a short id. |
/open <id or title> | Open a conversation — by its short id from the drawer or the list, or by the beginning of its title. |
/rename <new title> | Rename the current conversation. |
/delete <id or title> | Delete a conversation. |
/pin / /unpin | Keep the current conversation at the top of the list, or stop doing that. |
Long-term memory
| Command | What it does |
|---|---|
/memories | List what Ora remembers across conversations. |
/memories --scope <scope> | Narrow it: global, project:<name>, or service:<name>. |
/forget <memory-id> | Delete one entry. |
/forget --scope <scope> | Clear a whole scope. |
Memory is Ora's, not yours: it stores facts it learned while working — a contract detail, a quirk of an API, a decision you made — and reads them back into later conversations. The memories pill at the bottom of the panel lists the entries that were actually used in this conversation, and lets you delete one from there.
Examples
/plan Turn this specification into a data-driven regression project covering every endpoint
/thinking high
/model
/compact
/tokens
/memories --scope project:hummingbird-airline