Humanloop MCP. Manage prompt versions and deployment history directly in your chat.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Humanloop (LLM Prompt Management API) allows your AI agent to manage, version, and deploy LLM prompts directly. You can list all prompts, check version history, update metadata, and set specific versions as active in environments like staging or production.
It also lets you record model generations for evaluation and stream prompt execution.
What your AI agents can do
Call prompt stream
Runs a prompt and streams the resulting text output.
Delete prompt version
Removes a specific, obsolete version of a prompt from the library.
Deploy prompt
Sets a specific prompt version as the active deployment for a chosen environment (e.g., production or staging).
Runs a prompt and streams the resulting text output back to your agent.
Adds new prompts to your library or modifies existing ones with upsert_prompt.
Retrieves the full version history of a prompt or updates its name and description using list_prompt_versions and update_prompt_version.
Deploys a specific prompt version to a target environment (like production) using deploy_prompt.
Lists all environments and which prompt version is active in each one via list_prompt_environments.
Logs a model generation against a specific prompt ID for later evaluation and quality review.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
019e5d24call prompt stream
Runs a prompt and streams the resulting text output.
019e5d24delete prompt version
Removes a specific, obsolete version of a prompt from the library.
019e5d24deploy prompt
Sets a specific prompt version as the active deployment for a chosen environment (e.g., production or staging).
019e5d24get prompt
Retrieves all configuration details for a single prompt using its unique ID.
019e5d24list prompt environments
Shows all deployment environments and which prompt version is currently active in each one.
019e5d24list prompt versions
Gets a list of all historical versions associated with a specific prompt ID.
019e5d24list prompts
Retrieves a complete list of every prompt stored in your entire organization.
019e5d24log to prompt
Records a model generation (log) tied to a specific prompt ID for later evaluation.
019e5d24remove deployment
Deactivates a specific prompt version from a given environment.
019e5d24update monitoring
Activates or deactivates monitoring tools (Evaluators) for a prompt's log data.
019e5d24update prompt version
Changes the name or description of an existing prompt version record.
019e5d24upsert prompt
Creates a new prompt configuration or updates an existing one.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Humanloop (LLM Prompt Management API), then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
Humanloop API MCP Server - Prompt Version Control
Listen up. This server hooks your AI agent right into Humanloop so you can handle your whole prompt lifecycle—from drafting it to putting it into production—all without leaving your chat window. You don't gotta jump into a dashboard to manage your LLM prompts; your agent does it for you.
Managing the Prompt Library
Your agent lets you see every single prompt you've got stored in your organization using list_prompts. You can also pull all the setup details for one specific prompt ID by calling get_prompt. Need to add a new prompt or tweak an existing one? Use upsert_prompt to create or modify the configuration.
You can keep an eye on all the different deployment environments and see exactly which prompt version is running in each one by calling list_prompt_environments. **
**
Controlling Versions and Deployments
It's all about control. You can grab a full history of versions for any prompt ID using list_prompt_versions. You can also change the name or description of a version record with update_prompt_version. When you're ready, your agent sets a specific prompt version as the active deployment for a target environment—like production or staging—by running deploy_prompt.
If you need to take a version offline, remove_deployment deactivates it from a given environment. And if a version is trash, delete_prompt_version wipes it out completely from the library.
Running, Logging, and Auditing
Your agent runs the prompt and streams the resulting text output right back to you using call_prompt_stream. You can log model generations against a specific prompt ID using log_to_prompt, which is crucial for later evaluation and quality checks. For monitoring, you can activate or deactivate monitoring tools (Evaluators) for a prompt's log data with update_monitoring.
You'll get the full picture of where your prompts are deployed and what versions are active by calling list_prompt_environments.
It's that simple. Your agent handles all the heavy lifting, keeping your prompts clean, controlled, and ready to rock.
How Humanloop MCP Works
- 1 Subscribe to the server and provide your Humanloop API Key.
- 2 Call a tool like
list_promptsto see what prompts exist in your organization. - 3 Use
deploy_promptto promote a specific version to a target environment (e.g., production).
The bottom line is, you treat your prompt library like a version-controlled microservice, managing deployment and audit trails without leaving your agent's chat window.
Who Is Humanloop MCP For?
The AI Engineer who gets tired of context switching between the code editor and the prompt dashboard. The Product Manager who needs to audit prompt versions and monitor model output before a release. DevOps teams automating prompt configuration across multiple environments.
Tests and iterates prompt versions directly from the editor, calling tools like upsert_prompt and call_prompt_stream without leaving their IDE.
Audits prompt version history using list_prompt_versions and monitors model output via log_to_prompt to ensure quality before launch.
Automates the deployment of prompt configurations across staging and production using deploy_prompt and remove_deployment.
What Changes When You Connect
- Stop clicking through dashboards to manage prompts. Use
upsert_promptorlist_promptsto create, modify, or list every prompt in one conversation. - Audit your prompt history instantly.
list_prompt_versionsgives you a full timeline of changes, so you know exactly what version is running where. - Control releases with confidence. Use
deploy_promptto make a specific version live in production, andremove_deploymentto quickly roll back if something breaks. - Keep track of model performance.
log_to_promptrecords every generation, allowing you to evaluate the output quality without running manual tests. - See the deployment status at a glance.
list_prompt_environmentsshows which version is active across staging and production, saving you audit time. - Test prompts live. You can run prompts and stream the output immediately using
call_prompt_stream.
Real-World Use Cases
The QA Team needs to test a prompt change.
A QA engineer updates a prompt using upsert_prompt to fix a hallucination issue. They immediately run the new version using call_prompt_stream to confirm the fix, then use list_prompt_environments to check if the staging environment is pointing to the right version before going live.
A Product Manager needs to rollback a bad deployment.
The production prompt generates bad responses. The PM immediately calls list_prompt_versions to find the last known good version ID. They then use deploy_prompt to revert the entire environment to that stable version.
A DevOps Team needs to automate release governance.
The team writes a script that calls list_prompts to gather all necessary IDs. It then uses deploy_prompt for staging, waits for monitoring checks via update_monitoring, and finally promotes it to production.
A Researcher needs to track model performance.
A researcher wants to benchmark two different prompt versions. They run the prompt twice, calling log_to_prompt for both runs, and then use get_prompt to pull the full configuration for comparison.
The Tradeoffs
Treating prompts like static files
Manually updating the prompt text in a spreadsheet and copying it into the production system. This loses version history, makes rollbacks impossible, and forces the team to remember which version was actually deployed.
→
Use upsert_prompt to create the new version, followed by update_prompt_version to rename it. Then, use deploy_prompt to make that specific version active in the target environment. Always check list_prompt_versions first.
Forgetting to check the deployment status
Assuming a prompt change is live in production because the developer ran it locally. The actual production environment might still be pointing to an older, unretired version.
→
Always check list_prompt_environments to confirm the exact active version ID in every environment before declaring a release complete.
Overwriting prompts without logging the reason
Running upsert_prompt with new text and assuming it's fine. Without logging, there's no record of why the change was made or who approved it.
→
Before running upsert_prompt, use get_prompt to verify the current config. After the change, use log_to_prompt to record the execution context and rationale.
When It Fits, When It Doesn't
Use this if you need to treat your prompt library like critical, governed infrastructure. You need to track who changed what, when it was changed, and which specific version is running in production. Use it when the risk of a bad prompt going live is high. Don't use it if you are just doing quick, isolated experiments; simpler local dev tools are fine. If your primary goal is only to run the prompt and get text back, just use a basic chat tool. But if you need deployment guarantees, use deploy_prompt and list_prompt_environments.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Humanloop. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 12 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Changing a prompt shouldn't mean rebuilding the whole system.
Today, changing a core prompt requires jumping between documentation, the prompt dashboard, and the actual deployment system. You modify the text, then you manually check if the staging environment sees the change, and finally, you submit a ticket to deploy it to production. It's slow, and version control is a nightmare.
With the Humanloop API, you manage the entire cycle from your agent. You use `upsert_prompt` to update the text, then `deploy_prompt` to move it to staging. You can audit the whole thing, knowing exactly which version is active, and do it all from one chat session.
Humanloop (LLM Prompt Management API) MCP Server
You eliminate the manual process of checking environment status and version tags. Instead of clicking 'Deploy to Staging' and then 'Deploy to Production' across different UIs, you send a single command to deploy the desired version to multiple environments in sequence.
This means your prompt logic is treated as code—versioned, audited, and deployable—without touching a dashboard. It’s a direct, programmatic path from idea to production.
Common Questions About Humanloop MCP
How do I check what versions of a prompt are available using list_prompt_versions? +
Run list_prompt_versions and provide the prompt ID. This returns a list of all historical versions, including their version IDs and creation dates. It's how you find the last known good version for rollback.
Can I test a prompt using call_prompt_stream before deploying it? +
Yes. Use call_prompt_stream with the prompt ID. This executes the prompt and streams the result directly to your agent, letting you test the output before committing to a deployment.
What is the difference between list_prompts and list_prompt_environments? +
list_prompts gives you a list of every prompt in your organization. list_prompt_environments shows which specific version of a prompt is currently live and active in defined environments (like production).
How do I deploy a prompt version to production using deploy_prompt? +
You use deploy_prompt and specify the version ID and the target environment ('production'). The server handles the deployment and sets that version as the live one.
How do I set a prompt version as active in a specific environment using deploy_prompt? +
You use deploy_prompt to activate a version. This tool takes the target environment (like staging or production) and the specific prompt version ID. It ensures that the specified version is the one the system will use for all subsequent calls in that environment.
What is the purpose of log_to_prompt? +
It records model generations. When you use log_to_prompt, you associate a specific model output (the log) with a given prompt ID. This lets you audit performance and track how the prompt behaved over time.
How do I update the name or description of an existing prompt version using update_prompt_version? +
You call update_prompt_version to change metadata. This tool requires the prompt version ID, and you pass in the new name and description. It changes the label, not the underlying prompt content.
What information can I get about a prompt's current deployment status with list_prompt_environments? +
It lists all environments and shows which version is currently active for a given prompt. This helps you quickly see if 'production' is running the version you intended.
Can I see the full version history of a specific prompt? +
Yes! Use the list_prompt_versions tool with the Prompt ID. It will return all historical versions, allowing you to track changes and metadata over time.
How do I deploy a prompt version to a specific environment like production? +
You can use the deploy_prompt tool. Provide the Prompt ID and the Environment ID to set that specific version as the active deployment for that environment.
Is it possible to record model outputs for later evaluation? +
Absolutely. Use the log_to_prompt tool to record a generation, including the prompt path, messages, and output, which can then be used for evaluation in Humanloop.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
TOML Parser Engine
Convert TOML config files to JSON and back. Essential for Rust (Cargo.toml), Python (pyproject.toml), and Cloudflare (wrangler.toml).
BugHerd
Manage visual feedback and bug reports via BugHerd — track projects, tasks, and users directly from any AI agent.
Novu
Automate multi-channel notifications via Novu — trigger workflows, manage subscribers, and handle preferences directly from any AI agent.
You might also like
NationBuilder
Organize communities, run campaigns, and mobilize supporters with a platform built for political and nonprofit leadership.
VWO
Manage A/B tests, feature flags, and conversion goals on VWO — the leading experience optimization platform.
Elai AI Video
Equip your AI agent to generate AI videos, manage avatars, and track rendering status via the Elai.io API.