Agentic Reinforcement Learning

by hoyant-suco.huggingface.hoyant-su/agentic-rl

Credit assignment research and CLI agent benchmark tasks

  • Remote endpoint
  • 8 tools
  • CC-BY-4.0

Uptime

Checked  · every day

100%from 4 days of checks

Up without downtime on 4 of 4 days with data.

  • No downtime
  • Degraded
  • Down over 5 minutes
  • No data
  • Not checked yet

Response time

Median 911 ms, slowest 5% 933 ms for the handshake and the tool list, over 30 days.

1000 ms500 ms0

How far each check got

  • Reached the address 100%
  • Completed the handshake 100%
  • Listed its tools 100%

8 tools at the last answer.

Every day
DayResultChecksMedian
3 Nov (today)No downtime1907 ms
2 NovNo downtime1of 1916 ms
1 NovNo downtime1of 1927 ms
31 OctNo downtime1924 ms

A failure counts as down time only when a second check five minutes later fails too. How we check

Connect to the publisher's own endpoint

Add to ~/.config/agenvoy/mcp.json:

{
  "servers": {
    "agentic-rl": {
      "url": "https://hoyant-su-agentic-rl.hf.space/gradio_api/mcp/sse"
    }
  }
}

This server's endpoint speaks only SSE, which this client's documentation does not list among its transports; use another client for it.

For agents: the install prompt

Every step above as one Markdown text, for an assistant installing it for you.

# Install the MCP server Agentic Reinforcement Learning (`co.huggingface.hoyant-su/agentic-rl`)

Its page: https://plug.directory/servers/co.huggingface.hoyant-su/agentic-rl

It runs on the publisher's servers; the client connects to its URL.

Use the section for the person's client.

## Claude Desktop

Paste into Settings › Connectors › Add custom connector:

```text
https://hoyant-su-agentic-rl.hf.space/gradio_api/mcp/sse
```

## Claude Code

Run in a terminal:

```sh
claude mcp add --transport sse agentic-rl https://hoyant-su-agentic-rl.hf.space/gradio_api/mcp/sse
```

## Cursor, Devin Desktop, LM Studio, Gemini CLI, oterm, ECA

Merge into the place for the person's client:

- Cursor: ~/.cursor/mcp.json
- Devin Desktop: ~/.config/devin/mcp_config.json
- LM Studio: mcp.json (Program › Install › Edit mcp.json)
- Gemini CLI: ~/.gemini/settings.json
- oterm: config.json in oterm's data folder (~/Library/Application Support/oterm on macOS, ~/.local/share/oterm on Linux, %APPDATA%\oterm on Windows)
- ECA: ~/.config/eca/config.json (global) or .eca/config.json in the project

```json
{
  "mcpServers": {
    "agentic-rl": {
      "url": "https://hoyant-su-agentic-rl.hf.space/gradio_api/mcp/sse"
    }
  }
}
```

## VS Code

Merge into .vscode/mcp.json:

```json
{
  "servers": {
    "agentic-rl": {
      "type": "sse",
      "url": "https://hoyant-su-agentic-rl.hf.space/gradio_api/mcp/sse"
    }
  }
}
```

## Zed

This server's endpoint speaks only SSE, which this client's documentation does not list among its transports; use another client for it.

## Cline, AnythingLLM, Augment Code

Merge into the place for the person's client:

- Cline: cline_mcp_settings.json (MCP Servers › Configure › Configure MCP Servers)
- AnythingLLM: anythingllm_mcp_servers.json in the AnythingLLM storage plugins directory
- Augment Code: Augment Settings Panel – MCP section – Import from JSON (VS Code extension)

```json
{
  "mcpServers": {
    "agentic-rl": {
      "type": "sse",
      "url": "https://hoyant-su-agentic-rl.hf.space/gradio_api/mcp/sse"
    }
  }
}
```

## Continue

Merge into ~/.continue/config.yaml:

```yaml
mcpServers:
  - name: "agentic-rl"
    type: sse
    url: "https://hoyant-su-agentic-rl.hf.space/gradio_api/mcp/sse"
```

## Goose

This server's endpoint speaks only SSE, which this client's documentation does not list among its transports; use another client for it.

## Codex

This server's endpoint speaks only SSE, which this client's documentation does not list among its transports; use another client for it.

## mcp-client-cli, TUUI, BoltAI, Bob Shell, Kiro, Kiro CLI

This server's endpoint speaks only SSE, which this client's documentation does not list among its transports; use another client for it.

## Agenvoy

This server's endpoint speaks only SSE, which this client's documentation does not list among its transports; use another client for it.

## Amp

Merge into ~/.config/amp/settings.json (user) or .amp/settings.json (workspace):

```json
{
  "amp.mcpServers": {
    "agentic-rl": {
      "url": "https://hoyant-su-agentic-rl.hf.space/gradio_api/mcp/sse"
    }
  }
}
```

Then restart the client, or reload its MCP servers, and ask it to list the server's tools to confirm it started.

Tools 8

  • Agentic_RL_dataset_overview Inspect ShellOps and ShellOps-Pro task counts, train/test splits, task types, published schemas, source files, license and citation.
  • Agentic_RL_fetch_evidence Fetch a complete original evidence block by the evidence_id returned from search_evidence, including section anchor, version, equations, table cells, links, and attribution.
  • Agentic_RL_filter_methods Filter agent RL credit-assignment methods by research conditions and return original section evidence and BibTeX. Discover accepted values with list_method_facets. Filters combine with AND; empty strings leave a facet unrestricted. Unknown critic status never matches no. Results use publication order without a relevance or quality ranking.
  • Agentic_RL_get_task Inspect one published ShellOps or ShellOps-Pro task by its exact task_id and partition ('shellops' or 'shellops_pro'). Returns the complete instruction, actual reward specification, published reference answer/command, file-entry metadata, pinned parquet rows and workspace asset links. File content is available at the source links. No shell execution or solution verification is performed.
  • Agentic_RL_list_method_facets List exact filter values for agent RL credit granularity, supervision, value critics and evaluation settings. Each value reports its source-supported method count.
  • Agentic_RL_list_sources List original papers and retrieval coverage. Discover source-linked comparisons of credit assignment, agent memory, selective observation and terminal benchmarks, with JSON, CSV and BibTeX links.
  • Agentic_RL_search_evidence Search original papers on agentic reinforcement learning, credit assignment and CLI agents. Use English keywords (AND), OR and quoted phrases. Return relevant passages, source citations, equations and table cells.
  • Agentic_RL_search_tasks Find real ShellOps CLI benchmark tasks by case-insensitive literal substring in the complete instruction, task ID or published task type. Empty query lists all tasks. Select partition 'all', 'shellops' or 'shellops_pro'; select published split 'all', 'train_src', 'train' or 'test'. Results are ordered by partition then task ID, with explicit pagination and no relevance scoring. The train subset is not double-counted.

From Hugging Face Spaces. Report a problem