Unable to Run Playbooks with Xata Agent – Assistance Required

[Karthikr2212]

Hello Xata Team,

I’ve successfully installed the Xata Agent on one of our development PostgreSQL servers. The agent appears to be connecting to the database and collecting relevant information as expected.

However, I’m currently facing an issue when attempting to run playbooks — they fail to execute and throw an error. I’ve reviewed the Docker logs and noticed the following error message:

xata-agent-1 | responseBody: '{\n' +
xata-agent-1 | ' "error": {\n' +
xata-agent-1 | ' "message": "You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platfo
rm.openai.com/docs/guides/error-codes/api-errors.",\n' +
xata-agent-1 | ' "type": "insufficient_quota",\n' +
xata-agent-1 | ' "param": null,\n' +
xata-agent-1 | ' "code": "insufficient_quota"\n' +
xata-agent-1 | ' }\n' +
xata-agent-1 | '}\n',
xata-agent-1 | isRetryable: true,
xata-agent-1 | data: { error: [Object] }

Could you please assist in identifying the cause and guiding me on how to resolve this?

Best regards,
Karthik
PostgreSQL DBA
COPART

[Karthikr2212]

I am using opensource xata agent

[gulcin]

Hello @Karthikr2212, thanks for the detailed report and log excerpt.

The error in your Docker logs shows that the Xata Agent is trying to run playbooks requiring OpenAI access, but your account is out of quota/credits. Please check your OpenAI account usage and quota status. If exceeded, you’ll need to wait for reset or upgrade your plan.

Also confirm that OPENAI_API_KEY is correctly set in env.production file.

If you’ve restored quota or added a valid key and still see the issue, let us know and we’ll assist further.

[Karthikr2212]

Hello Xata Team,

Is the OPENAI_API_KEY required in env.production? I used a test key, but I'm seeing the error: "You exceeded your current quota, please check your plan and billing details." It seems like I need a paid API key. Is there any way to run the playbooks using a test key?

Regards,
Karthik
PostgreSQL DBA
COPART

[gulcin]

Hello @Karthikr2212, you shouldn’t need an OpenAI API key for using Agent. Could you please share which Agent version you’re using?

[bojannamk1]

Can we use this agent without open API key ?

[gulcin]

Can we use this agent without open API key ?

Yes.

[bojannamk1]

Can we use this agent without open API key ?

Yes.

Thank you so much. I will implement this. Very promising and excited to use this !

[gulcin]

Can we use this agent without open API key ?

Yes.

Thank you so much. I will implement this. Very promising and excited to use this !

Glad to hear it, let us know if there are any problems!

[Karthikr2212]

Hi @gulcin ,

I am using xataio/agent:0.3.1 version, I am able to connect DB ,but unable to use playbooks. It throwing errors

[Karthikr2212]

without API key i havent seen playbooks and chat option. Getting below error in docker messages and getting 500 | internal server

Error: No providers enabled. Please configure API keys

[divyenduz]

Thanks for reaching out again, the error you mention comes from this code path here. Please double check that the names of the env vars are correct (typo or anything?) and if they are being loaded into docker correctly.

If the problem persists, please share reproduction steps of your setup. We will try to then reproduce it on our end.

[Karthikr2212]

PUBLIC_URL=http://10.x.x.x:8080
AUTH_SECRET=KsbjBuAKpkW4+G2AJMXYSlEln8BOA3G1wcWkDRzOjj8=
POSTGRES_HOST=10.x.x.x
POSTGRES_PORT=5432
POSTGRES_USER=xata_moni_usr
POSTGRES_PASSWORD=xxxxxxxx
POSTGRES_DB=yds

i used above fields in .env.production and got the below error

[✓] migrations applied successfully!Starting scheduler with 60s interval (60000ms)
xata-agent-1 | ▲ Next.js 15.3.2
xata-agent-1 | - Local: http://localhost:8080
xata-agent-1 | - Network: http://172.18.0.3:8080
xata-agent-1 |
xata-agent-1 | ✓ Starting...
xata-agent-1 | ✓ Ready in 1358ms
xata-agent-1 | Error: No providers enabled. Please configure API keys
xata-agent-1 | at (.next/server/chunks/4855.js:1:2298)
xata-agent-1 | at 34530 (.next/server/chunks/4855.js:1:2509)
xata-agent-1 | at t (.next/server/webpack-runtime.js:1:143)
xata-agent-1 | at 86327 (.next/server/chunks/4855.js:190:2859)
xata-agent-1 | at t (.next/server/webpack-runtime.js:1:143)
xata-agent-1 | at (.next/server/app/api/chat/route.js:1:17767)
xata-agent-1 | at t.a (.next/server/webpack-runtime.js:1:918)
xata-agent-1 | at 97382 (.next/server/app/api/chat/route.js:1:17680)
xata-agent-1 | at t (.next/server/webpack-runtime.js:1:143)
xata-agent-1 | at (.next/server/app/api/chat/route.js:1:3750)
xata-agent-1 | ⨯ unhandledRejection: Error: No providers enabled. Please configure API keys
xata-agent-1 | at (.next/server/chunks/4855.js:1:2298)
xata-agent-1 | at 34530 (.next/server/chunks/4855.js:1:2509)

[Karthikr2212]

if the above configuration is not correct, just give me the sample configuration file

[divyenduz]

Ah, you need to add one of the AI provider API keys, so one of OPEN_AI_API_KEY, ANTHROPIC_API_KEY and so on should exist. In the config you shared above, there is no AI provider API key set.

[Karthikr2212]

Yes thats what i am saying without API key set playbooks and chat option not available
Do we need to purchase the OPEN_AI_API_KEY subscription ?

[Karthikr2212]

[Karthikr2212]

it means only we can use to connect database and collecting environment variables information only ?

[divyenduz]

Yes, there has to be at least one AI provider for the chat to work, the chat is powered by LLMs, without it, it does not work. Sorry about the confusion earlier.

[bojannamk1]

Ok so without API key it doesn't work :( I was getting be same error.

[bojannamk1]

Can we get something to use the offline models like llama 3.2 until we purchase the keys? Thanks in advance.

[Karthikr2212]

Hi Team,

Can we use our own organization internal api keys(https://testapi.copart.com/api), If that works please guide me how to implement

[divyenduz]

Yes, it works with local LLMs,

agent/apps/dbagent/src/lib/env/server.ts

Lines 29 to 30 in 30aecbc

	OLLAMA_HOST: z.string().optional(),
	OLLAMA_HEADERS: z.record(z.string(), z.string()).optional(),

You can provide OLLAMA_HOST env var to do so.

[divyenduz]

own organization internal api keys

what's the setup like, that URL is not publicly accessible, maybe there is a way to make it work with LiteLLM which agent support but I am not yet sure on what is on your side of API.

[Karthikr2212]

Can you please give me a example that what i have to update
below is the our oranization api

https://gegie.copart.com

[divyenduz]

It is a private URL, likely only accessible behind a VPN.

[Karthikr2212]

Actually this is my organization internal one(we can get api keys)...so it cannot be access outside

So can you please let me know what xchanges need to be done in below file

[divyenduz]

Each model uses a different API, depending on which one is at your org, you would need to choose one of Ollama or LiteLLM. I do not have enough information to help you without knowing what flavor of AI API is offered internally at your org. Maybe try asking ChatGPT or someone at your org who has setup the internal one?

[aryaanbomma]

@divyenduz is there a way we could set a custom OpenAI base URL? Maybe an environment variable we could configure like so: OPENAI_API_BASE=<custom_base_url>

[divyenduz]

I think currently it is only possible via a proxy (via LiteLLM, and ollama can do that too if I am not mistaken). But might be a good candidate for a feature request. Mind opening one?

[Karthikr2212]

Yes Sure, Can you please help me to raise a feature request. How to raise it?

[Karthikr2212]

Any update ?

[divyenduz]

https://github.com/xataio/agent/issues/new

You can make one here, write your use case in detail and how you would imagine the feature would work.

[divyenduz]

To summarize, we need at least one AI model to make the chat work. It doesn't need to be OpenAI but can be Anthropic or something else. Conceptually, there are 3 types of models

1. built-in - Open AI, Anthropic etc.
2. ollama - self hosted models via ollama
3. LLMLite - proxy layer that converts models to open AI spec

When one of them is available the agent chat is functional.