Integration · Developer tools

Connect Claude & ChatGPT to Keywords AI

Ask your AI what your LLM app did yesterday, and it reads the request logs, traces and usage totals from Keywords AI, now called Respan. It can open a single request with its prompt, response, tokens and cost, and see usage per end customer. It can also create prompts and new prompt versions, deploy a version, and build datasets from your logs.

Connect Keywords AI free

Free plan, no credit card. Takes about a minute.

Try asking

“How many requests, tokens and dollars did we use in the last 24 hours?”
The AI reads the usage summary for that period with total requests, tokens, cost and latency.
“Show me the failed requests from production this morning.”
It lists request logs in that time range for the production environment, filtered on status, with model and cost.
“Open log abc123 and show what the model answered.”
It reads that request log with the prompt, the response, the model, tokens, cost and latency.

What your AI can do

Logs, traces and usage

List and read request logs and traces, and get totals for requests, tokens, cost and latency over any period.

Prompts

List and read managed prompts and their versions, create prompts and versions, and deploy a version to production.

Datasets and evaluations

Create datasets from logs, and read datasets, testsets, evaluators, experiments and end customers.

“Which of our customers used the most this month?”
It lists your end customers with their usage and can open one customer to see cost and budget.
“Walk me through the trace for last night's failed agent run.”
It finds the trace in the time range and reads all its spans in order.
“Add a version of the support-reply prompt with temperature 0.2, but don't deploy it yet.”
It creates a new prompt version with the messages, model and settings you give. The live version stays the same until you deploy the new one.

27 tools for Keywords AI

These tools are switched on when you connect. You can switch any of them off, or require your approval before it runs.

create_datasetChanges dataCreate a dataset with `name` and `description`, either empty (`is_empty`) or sampled from logs between `start_time` and `end_time`, or copied from `source_dataset_id`.
create_promptChanges dataCreate a new managed prompt with a `name` and optional `description`. Add content with create_prompt_version.
create_prompt_versionChanges dataCreate a new version of a prompt by `prompt_id` with `messages`, `model`, `temperature` and other settings. `deploy` true makes it live immediately.
deploy_prompt_versionChanges dataMake a prompt `version` the live version for `prompt_id`. Production calls using this prompt switch to it right away.
get_customerGet one end customer by `customer_identifier`: usage, cost and budget.
get_datasetGet one dataset by `dataset_id`.
get_evaluatorGet one evaluator by `evaluator_id` with its configuration.
get_experimentGet one experiment by `experiment_id` with its status and results.
get_llm_metrics_summaryChanges dataGet dashboard LLM usage metrics for a period, using `summary_type` (daily, weekly, monthly, last_24h and more) or `start_time`/`end_time`, with optional filters.
get_promptGet one prompt by `prompt_id`, including its current version.
get_prompt_versionGet one version of a prompt by `prompt_id` and `version`: messages, model and settings.
get_request_logGet one request log by `unique_id`: prompt, response, model, tokens, cost and latency.
get_request_logs_summaryChanges dataGet totals (requests, tokens, cost, latency) for request logs in a time range, with optional `filters` and `environment`.
get_testsetGet one testset by `testset_id`.
get_traceGet one trace by `trace_unique_id` with all its spans.
list_customersChanges dataList your end customers (the users of your AI app) with their usage, optionally filtered by `environment`.
list_dataset_logsChanges dataList the logs in a dataset by `dataset_id`, with optional filters.
list_datasetsChanges dataList datasets, with optional filters and sorting.
list_evaluatorsChanges dataList evaluators, with optional filters and sorting.
list_experimentsChanges dataList experiments, with optional filters and sorting.
list_prompt_versionsList the versions of a prompt by `prompt_id`.
list_promptsChanges dataList managed prompts, optionally filtered by name, tags or dates. Use to find a prompt_id.
list_request_logsChanges dataList LLM request logs, filtered by `start_time`, `end_time`, `environment`, `log_type` and `filters` (e.g. model, status, cost, customer). Use to find a log's unique_id.
list_testset_rowsList the rows of a testset by `testset_id`.
list_testsetsChanges dataList testsets, with optional filters and sorting.
list_tracesChanges dataList traces (multi-step agent or workflow runs) in a time range, with optional `filters` and sorting.
update_promptChanges dataRename a prompt or change its description by `prompt_id`.

Set up in three steps

1
Pick the app
Create a free PipMCP account and choose Keywords AI from the app list.
2
Paste your key
Paste a Respan (Keywords AI) API key. Keywords AI is now called Respan: log in at platform.respan.ai, open Settings > API keys, click Create new key, choose the environment and copy the key when it is shown.
3
Add the link to your AI
You get a personal MCP link. Add it to Claude, ChatGPT or Cursor:
1. Click your name, then Settings › Connectors › Add custom connector.
2. Paste your link as the Remote MCP server URL.
3. Switch it on from the + menu in a chat.

Questions

What can the AI do in Keywords AI?

The AI can list and read request logs, traces, prompts, prompt versions, datasets, testsets, evaluators, experiments and end customers, and get usage and metrics summaries. It can create datasets, prompts and prompt versions, rename a prompt, and deploy a prompt version so it becomes live.

Does the AI see my Keywords AI credentials?

No. Your Keywords AI credentials are encrypted at rest and never shown to the AI. After you save them, they are not shown again, not even to you. The AI only sees the results of the tools it calls.

Can I control what the AI is allowed to do?

Yes. You choose which tools are switched on, so you can start with only reading logs and usage. Deploying a prompt version switches production calls that use the prompt right away, so it can require your approval before it runs. Every tool call is logged.

Does it work with ChatGPT?

Yes. In ChatGPT go to Settings › Apps & Connectors › Advanced and turn on Developer mode, then add your PipMCP link. Developer mode needs a paid ChatGPT plan: Plus, Pro, Business or Enterprise. The same link also works in Claude (Settings › Connectors › Add custom connector), Cursor and other MCP clients.

What does it cost?

PipMCP has a free plan with no credit card required. Paid plans bill per completed task. You also need your own Keywords AI (Respan) account.

Keywords AI is now called Respan. Does this still work?

Yes. It connects to the same account under the new name. You create the API key at platform.respan.ai under Settings > API keys, and choose which environment the key belongs to when you create it.