feat(aiservices): add AI API Key management

[loicrouillermonay]

This PR adds CLI commands for managing AI API keys under exo ai-api-key:

• create --name <name> --scope <public|deployment-uuid> - creates a new API key and prints the plaintext value
• list - displays a table with id, name, scope, and created-at
• get <id> - shows detailed view (without the value field)
• update <id> [--name <name>] [--scope <scope>] - updates an existing API key
• delete <id> - deletes an API key
• rotate <id> - rotates an API key and prints the new plaintext value
• reveal <id> - reveals the plaintext value of an existing API key

This PR is marked as draft, waiting for a dependency merge in egoscale.

[arnaudgeiser]

Wouldn't it make sense to actually create ai subcommands? That would match the experience we have on the portal.
But I do think that dedicated-inference should be moved there as-well.

[loicrouillermonay]

Wouldn't it make sense to actually create ai subcommands? That would match the experience we have on the portal. But I do think that dedicated-inference should be moved there as-well.

cc @anamitev

Right now it's exo dedicated-inference ... but exo ai ... is an alias working the same way (that I'm already always using because it's shorter).

For the next XIP about "Managed Inference GA" (not MVP) we could consider making exo ai the main one and dedicated-inference an alias of it because it will be a weird UX to create a Managed Inference key via exo dedicated-inference ai-api-key create

[retrack]

Yes that would respect the umbrella concept as presented to customers and consistent with dbaas for example. In the absence of Ana, ok from us.

…

-- Antoine Le lun. 6 juil. 2026 à 14:18, Loïc Rouiller-Monay ***@***.***> a écrit :

*loicrouillermonay* left a comment (exoscale/cli#828) <#828 (comment)> Wouldn't it make sense to actually create ai subcommands? That would match the experience we have on the portal. But I do think that dedicated-inference should be moved there as-well. cc @anamitev <https://github.com/anamitev> Right now it's exo dedicated-inference ... but exo ai ... is an alias working the same way (that I'm already always using because it's shorter). For the next XIP about "Managed Inference GA" (not MVP) we could consider making exo ai the main one and dedicated-inference an alias of it because it will be a weird UX to create a Managed Inference key via exo dedicated-inference ai-api-key create — Reply to this email directly, view it on GitHub <#828?email_source=notifications&email_token=AAJ4HKONZN5WVGLJUCTWXAT5DOKKDA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIOBZGI3DQNBZGY22M4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#issuecomment-4892684965>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJ4HKK7XF6HLHNHVEIMG2D5DOKKDAVCNFSNUABFKJSXA33TNF2G64TZHMYTIOJSGU3TQNZQHNEXG43VMU5TIMZVHEYTMMZSGM2KC5QC> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/AAJ4HKJXUSGYF55YFUSGAVD5DOKKDA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIOBZGI3DQNBZGY22M4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSVGM33PORSXEX3JN5ZQ> and Android <https://github.com/notifications/mobile/android/AAJ4HKM5G5NONYYI24KBA7T5DOKKDA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIOBZGI3DQNBZGY22M4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ>. Download it today! You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>