FreshRSS Agent

CLI or API | MCP | Agent

Version: 1.0.1

Documentation — Installation, deployment, usage across the API, CLI, and MCP interfaces, the integrated A2A agent server, and guidance for provisioning the backing platform are maintained in the official documentation.

---

Table of Contents

• Overview
• Key Features
• MCP
  • Available MCP Tools
  • Environment Variables
  • Transport Configuration Examples
  • Additional Deployment Options
• Usage
• Installation
• Documentation

---

Overview

FreshRSS MCP Server + A2A Agent

A connector for the self-hosted FreshRSS RSS reader, wrapping its Google Reader compatible API (GReader). It exposes two action-routed MCP tool domains:

• freshrss_reader — stream_contents (feed items + continuation), item_contents, unread_count.
• freshrss_subscriptions — list, subscribe, unsubscribe, label, categories, mark_read, star.

This repository is actively maintained - Contributions are welcome!

---

Key Features

• Consolidated Action-Routed MCP Tools: Two togglable tool domains group every GReader operation, minimizing token overhead and tool bloat in LLM contexts.
• Google Reader Compatible: Wraps the FreshRSS GReader API — ClientLogin auth, transparent re-authentication on 401, and automatic write-token handling.
• Enterprise-Grade Security: OIDC token delegation (RFC 8693), Eunomia policy enforcement, and per-instance credential resolution.
• Integrated A2A Agent: Built-in Pydantic AI agent server alongside the MCP server.
• Native Telemetry & Tracing: Out-of-the-box OpenTelemetry exports and Langfuse tracing.

---

MCP

Install the slim [mcp] extra. All MCP examples below install freshrss-agent[mcp] — the MCP-server extra that pulls only the FastMCP / FastAPI tooling (agent-utilities[mcp]). It deliberately excludes the heavy agent runtime (the epistemic-graph engine, pydantic-ai, dspy, llama-index, tree-sitter), so uvx/container installs are dramatically smaller and faster. Use the full [agent] extra only when you need the integrated Pydantic AI agent (see Installation).

Available MCP Tools

Auto-generated from the live MCP server — do not edit by hand.

Condensed action-routed tools (default — MCP_TOOL_MODE=condensed)

MCP Tool	Toggle Env Var	Description
freshrss_reader	READERTOOL	Read FreshRSS streams via the Google Reader API. CONCEPT:FR-OS.identity.frss
freshrss_subscriptions	SUBSCRIPTIONSTOOL	Curate FreshRSS feeds, categories and item tags. CONCEPT:FR-OS.governance.frss

Verbose 1:1 API-mapped tools (MCP_TOOL_MODE=verbose or both)

10 per-operation tools — one per public API method (click to expand)

MCP Tool	Toggle Env Var	Description
freshrss_categories	SUBSCRIPTIONS_MIXINTOOL	List categories / tags (tag/list).
freshrss_item_contents	READER_MIXINTOOL	Fetch full contents for specific item ids (GReader i parameters).
freshrss_label	SUBSCRIPTIONS_MIXINTOOL	Add a category label to an existing feed subscription.
freshrss_mark_read	SUBSCRIPTIONS_MIXINTOOL	Mark one or more items as read.
freshrss_star	SUBSCRIPTIONS_MIXINTOOL	Star or unstar an item.
freshrss_stream_contents	READER_MIXINTOOL	Fetch items for a stream.
freshrss_subscribe	SUBSCRIPTIONS_MIXINTOOL	Subscribe to a feed, optionally setting its title and category.
freshrss_subscription_list	SUBSCRIPTIONS_MIXINTOOL	List all feed subscriptions.
freshrss_unread_count	READER_MIXINTOOL	Return unread counts per stream.
freshrss_unsubscribe	SUBSCRIPTIONS_MIXINTOOL	Unsubscribe from a feed.

2 action-routed tool(s) (default) · 10 verbose 1:1 tool(s). Each is enabled unless its <DOMAIN>TOOL toggle is set false; MCP_TOOL_MODE selects the surface (condensed default · verbose 1:1 · both). Auto-generated — do not edit.

Detailed tool schemas, parameter shapes, and validation constraints are preserved in docs/usage.md.

Environment Variables

Package environment variables

Variable	Example	Description
HOST	0.0.0.0
PORT	8000
TRANSPORT	stdio	options: stdio, streamable-http, sse
ENABLE_OTEL	True
OTEL_EXPORTER_OTLP_ENDPOINT	http://localhost:8080/api/public/otel
OTEL_EXPORTER_OTLP_PUBLIC_KEY	pk-...
OTEL_EXPORTER_OTLP_SECRET_KEY	sk-...
OTEL_EXPORTER_OTLP_PROTOCOL	http/protobuf
EUNOMIA_TYPE	none	options: none, embedded, remote
EUNOMIA_POLICY_FILE	mcp_policies.json
EUNOMIA_REMOTE_URL	http://eunomia-server:8000
FRESHRSS_URL	http://localhost:8080
FRESHRSS_USER	admin
FRESHRSS_API_PASSWORD	your_api_password_here
FRESHRSS_SSL_VERIFY	True
READERTOOL	True
SUBSCRIPTIONSTOOL	True

Inherited agent-utilities variables (apply to every connector)

Variable	Example	Description
MCP_TOOL_MODE	condensed	Tool surface: condensed
MCP_ENABLED_TOOLS	—	Comma-separated tool allow-list
MCP_DISABLED_TOOLS	—	Comma-separated tool deny-list
MCP_ENABLED_TAGS	—	Comma-separated tag allow-list
MCP_DISABLED_TAGS	—	Comma-separated tag deny-list
MCP_CLIENT_AUTH	—	Outbound MCP auth (oidc-client-credentials for fleet calls)
OIDC_CLIENT_ID	—	OIDC client id (service-account auth)
OIDC_CLIENT_SECRET	—	OIDC client secret (service-account auth)
DEBUG	False	Verbose logging
PYTHONUNBUFFERED	1	Unbuffered stdout (recommended in containers)
MCP_URL	http://localhost:8000/mcp	URL of the MCP server the agent connects to
PROVIDER	openai	LLM provider for the agent
MODEL_ID	gpt-4o	Model id for the agent
ENABLE_WEB_UI	True	Serve the AG-UI web interface

17 package + 14 inherited variable(s). Auto-generated from .env.example + the shared agent-utilities set — do not edit.

Every variable the server reads. A copy-paste template lives in .env.example.

Connection & Credentials

Variable	Description	Default
FRESHRSS_URL	Base URL of the FreshRSS instance (e.g. http://freshrss.arpa)	http://localhost:8080
FRESHRSS_USER	FreshRSS username (GReader Email field)	—
FRESHRSS_API_PASSWORD	FreshRSS API password (Settings → Authentication)	—
FRESHRSS_SSL_VERIFY	Whether to verify TLS certificates	True

MCP server / transport

Variable	Description	Default
TRANSPORT	stdio, streamable-http, or sse	stdio
HOST	Bind host (HTTP transports)	0.0.0.0
PORT	Bind port (HTTP transports)	8000
MCP_TOOL_MODE	Tool surface: condensed, verbose, or both	condensed

Telemetry & governance

Variable	Description	Default
ENABLE_OTEL	Enable OpenTelemetry / Langfuse export	True
EUNOMIA_TYPE	Authorization mode: none, embedded, remote	none
EUNOMIA_POLICY_FILE	Embedded policy file	mcp_policies.json
EUNOMIA_REMOTE_URL	Remote Eunomia server URL	—

Tool toggles — each action-routed tool domain can be disabled via its toggle env var (set to false): READERTOOL, SUBSCRIPTIONSTOOL (see the Available MCP Tools table above).

MCP Configuration Examples

Install the slim [mcp] extra. All examples install freshrss-agent[mcp] — the MCP-server extra that pulls only the FastMCP / FastAPI tooling (agent-utilities[mcp]). It deliberately excludes the heavy agent runtime (pydantic-ai, the epistemic-graph engine, dspy, llama-index), so uvx / container installs are far smaller. Use the full [agent] extra only when you need the integrated Pydantic AI agent.

stdio Transport (local IDEs — Cursor, Claude Desktop, VS Code)

{
  "mcpServers": {
    "freshrss-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "freshrss-agent[mcp]",
        "freshrss-mcp"
      ],
      "env": {
        "MCP_TOOL_MODE": "condensed",
        "FRESHRSS_API_PASSWORD": "your_api_password_here",
        "FRESHRSS_URL": "http://localhost:8080",
        "FRESHRSS_USER": "admin",
        "READERTOOL": "True",
        "SUBSCRIPTIONSTOOL": "True"
      }
    }
  }
}

Streamable-HTTP Transport (networked / production)

{
  "mcpServers": {
    "freshrss-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "freshrss-agent[mcp]",
        "freshrss-mcp",
        "--transport",
        "streamable-http",
        "--port",
        "8000"
      ],
      "env": {
        "TRANSPORT": "streamable-http",
        "HOST": "0.0.0.0",
        "PORT": "8000",
        "MCP_TOOL_MODE": "condensed",
        "FRESHRSS_API_PASSWORD": "your_api_password_here",
        "FRESHRSS_URL": "http://localhost:8080",
        "FRESHRSS_USER": "admin",
        "READERTOOL": "True",
        "SUBSCRIPTIONSTOOL": "True"
      }
    }
  }
}

Alternatively, connect to a pre-deployed Streamable-HTTP instance by url:

{
  "mcpServers": {
    "freshrss-mcp": {
      "url": "http://localhost:8000/freshrss-mcp/mcp"
    }
  }
}

Deploying the Streamable-HTTP server via Docker:

docker run -d \
  --name freshrss-mcp-mcp \
  -p 8000:8000 \
  -e TRANSPORT=streamable-http \
  -e HOST=0.0.0.0 \
  -e PORT=8000 \
  -e MCP_TOOL_MODE=condensed \
  -e FRESHRSS_API_PASSWORD=your_api_password_here \
  -e FRESHRSS_URL=http://localhost:8080 \
  -e FRESHRSS_USER=admin \
  -e READERTOOL=True \
  -e SUBSCRIPTIONSTOOL=True \
  knucklessg1/freshrss-agent:mcp

Auto-generated from the code-read env surface (MCP_TOOL_MODE + package vars) — do not edit.

Additional Deployment Options

freshrss-agent can also run as a local container (Docker / Podman / uv) or be consumed from a remote deployment. The Deployment guide has full, copy-paste mcp_config.json for all four transports — stdio, streamable-http, local container / uv, and remote URL:

• Local container / uv — launch the server from mcp_config.json via uvx, docker run, or podman run, or point at a local streamable-http container by url.
• Remote URL — connect to a server deployed behind Caddy at http://freshrss-mcp.arpa/mcp using the "url" key.

Usage

Once configured, an LLM (or a direct caller) invokes a tool domain with an action and a JSON params_json payload. Examples:

// Fetch the 50 most recent unread items, newest first
{
  "tool": "freshrss_reader",
  "action": "stream_contents",
  "params_json": "{\"count\": 50, \"order\": \"n\"}"
}

// Subscribe to a feed and file it under a category
{
  "tool": "freshrss_subscriptions",
  "action": "subscribe",
  "params_json": "{\"feed_url\": \"http://example.com/rss\", \"category\": \"News\"}"
}

// Mark items as read
{
  "tool": "freshrss_subscriptions",
  "action": "mark_read",
  "params_json": "{\"item_ids\": [\"tag:google.com,2005:reader/item/0001\"]}"
}

Invoking a tool with an unknown or omitted action returns the discovery payload listing every valid action for that domain.

Installation

Pick the extra that matches what you want to run:

Extra	Installs	Use when
freshrss-agent[mcp]	Slim MCP server only (agent-utilities[mcp] — FastMCP/FastAPI)	You only run the MCP server (smallest install / image)
freshrss-agent[agent]	Full agent runtime (agent-utilities[agent,logfire] — Pydantic AI + the epistemic-graph engine)	You run the integrated agent
freshrss-agent[all]	Everything (mcp + agent + logfire)	Development / both surfaces

# MCP server only (recommended for tool hosting — slim deps)
uv pip install "freshrss-agent[mcp]"

# Full agent runtime (Pydantic AI + epistemic-graph engine)
uv pip install "freshrss-agent[agent]"

# Everything (development)
uv pip install "freshrss-agent[all]"      # or: python -m pip install "freshrss-agent[all]"

After installation two console scripts are available:

freshrss-mcp      # run the MCP server
freshrss-agent    # run the A2A agent server

Container images (:mcp vs :agent)

One multi-stage docker/Dockerfile builds two right-sized images, selected by --target:

Image tag	Build target	Contents	Entrypoint
knucklessg1/freshrss-agent:mcp	--target mcp	freshrss-agent[mcp] — slim, no engine/pydantic-ai/dspy/llama-index/tree-sitter	freshrss-mcp
knucklessg1/freshrss-agent:latest	--target agent (default)	freshrss-agent[agent] — full agent runtime + epistemic-graph engine	freshrss-agent

docker build --target mcp   -t knucklessg1/freshrss-agent:mcp    docker/   # slim MCP server
docker build --target agent -t knucklessg1/freshrss-agent:latest docker/   # full agent

docker/mcp.compose.yml runs the slim :mcp server; docker/agent.compose.yml runs the agent (:latest) with a co-located :mcp sidecar.

Knowledge-graph database (epistemic-graph)

The full agent ([agent] / :latest) embeds the epistemic-graph engine (pulled in transitively via agent-utilities[agent]). For production — or to share one knowledge graph across multiple agents — run epistemic-graph as its own database container and point the agent at it instead of embedding it. Deployment recipes (single-node + Raft HA), connection config, and the full database architecture (with diagrams) are documented in the epistemic-graph deployment guide. The slim [mcp] server does not require the database.

Documentation

Full installation, deployment, usage, and platform-provisioning guides live in the docs/ directory and are published via mkdocs + GitHub Pages at the official documentation site:

• Overview · Installation · Usage · Deployment · Platform · Concepts

Deploy with agent-os-genesis

This package can be provisioned for you — skill-guided — by the agent-os-genesis universal skill (its single-package deploy mode): it picks your install method, seeds secrets to OpenBao/Vault (or .env), trusts your enterprise CA, registers the MCP server, and verifies it — the same machinery that stands up the whole Agent OS, narrowed to just this package. Ask your agent to "deploy freshrss-agent with agent-os-genesis".

Install mode	Command
Bare-metal, prod (PyPI)	uvx freshrss-mcp · or uv tool install freshrss-agent
Bare-metal, dev (editable)	uv pip install -e ".[all]" · or pip install -e ".[all]"
Container, prod	deploy knucklessg1/freshrss-agent:latest via docker-compose / swarm / podman / podman-compose / kubernetes
Container, dev (editable)	deploy docker/compose.dev.yml (source-mounted at /src; edits live on restart)

Secrets are read-existing + seeded via vault_sync — you are only prompted for what's missing.