Bagel
Bagel is an MCP server for inspecting, querying, reducing, and exporting robotics/drone/IoT data, with live subscriptions and reusable workflow/pipeline management.
Describe data sources and topics, including metadata, schemas, and query instructions.
Query topic messages using read-only DuckDB SQL over time ranges.
Read textual INFO/WARN/ERROR logs from recorded sources.
Discover, subscribe to, and unsubscribe from live MQTT/ROS bridge topics, persisting messages for analysis.
List, run, save, and delete reusable POML/Markdown agent capabilities.
Discover pipeline building blocks; preview, save, list, inspect, delete, run, and batch-run event-driven reduction pipelines.
Preview anomaly detection over recorded logs (beta) and snapshot robot hardware state with waffle-iron (beta).
Export selected event windows for PlotJuggler, Rerun, Lichtblick/Foxglove, or as LeRobot training datasets (beta).
Adds support for InfluxDB 3, allowing natural-language queries on IoT time-series data.
Enables natural-language analysis of live MQTT streams (including Sparkplug B) for IoT data monitoring and event detection.
Enables querying data stored in PostgreSQL databases using natural language, with the ability to generate deterministic SQL queries for analysis.
Allows users to chat with ROS1/ROS2 bag files and ROS log files, supporting queries on topics, metadata, and sensor data using natural language.
Adds support for TimescaleDB, a time-series database, to query and analyze time-stamped IoT data via natural language.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Bagel What's the peak z-deceleration in /imu?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Bagel by Extelligence lets you ask questions about robotics, drone, and IoT data in plain English. Every calculation over your message data is DuckDB SQL, not model guesswork, and Bagel shows you the query so you can audit it.
Is my IMU sensor overheating?
Bagel also has an intelligent edge data reduction pipeline: describe an event and Bagel runs the detection on the robot, keeping the windows that matter and dropping the rest. An MCP server puts all of it in your LLM's hands: Claude Code, Gemini, Cursor, or a fully local model.
Bagel was the first MCP server to ship a real analysis toolkit for robotics data, and it keeps the LLM where it belongs: in front of your logs, never in your robot's control loop.
📦 Install
Bagel is on PyPI as bagel-mcp. No Docker
and no ROS install needed for recorded data.
Claude Code, one line (needs uv):
claude mcp add bagel -- uvx --from "bagel-mcp[ros]" bagel-mcp --transport stdiopip, into any Python 3.10 to 3.12 environment:
pip install "bagel-mcp[ros]"Then point your MCP client at the bagel-mcp command it installed (setup for
Claude Desktop, Cursor and Codex is in the Quickstart).
Open your agent and ask:
Summarize the bag "~/logs/run_42.mcap".
The ros extra reads ROS 1 and ROS 2 bags. Flight logs, CAN / MDF4, live MQTT
and the rest are extras too. Live ROS
robots (rosbridge) and fleet/edge pipelines run in Docker.
🥯 Key Features
Ask in plain language: No deep domain expertise needed.
Transparent calculations: Deterministic SQL queries. No black-box LLM math.
Natural-language pipelines: "Keep 10s around every hard brake, drop the rest": one sentence becomes an auditable pipeline: previewed before a byte is written, then run once, across a fleet, or standing at the edge.
Broad LLM support: Claude Code, Gemini, Cursor, Codex, Grok Build, Grok Bot, and more.
Installs your way:
pip/uvxfor recorded data, Docker images for live robots and the edge.Extensible capabilities: Bagel can learn new tricks.
Wide format coverage: Missing your data format? Open a ticket.
Related MCP server: Robotics MCP Server
🥯 Try it in 60 seconds
No MCP client, no LLM, no config: run the same deterministic checks against a bundled sample log and get a robot-health report card straight to your terminal.
docker run -it --rm ghcr.io/extelligence-ai/bagel/px4:latest demosample.ulg - 41.5s, 2018 messages, 77 topics
Power ⚠️ min 21.07V, largest drop 2.37V at ~t=+4.8s, end 23.45V (battery_status_0)
IMU ✅ accel_z stddev 1.6x the log baseline at ~t=+36.8s (sensor_combined_0)
GPS — skipped: no GPS topic
Data gaps ✅ no gap > 1.05x median interval (checked battery_status_0, sensor_combined_0)
...The ROS2 images (ros2-kilted, ros2-jazzy, ros2-iron, ros2-humble) run
demo the same way, against a lighter bundled sample (px4 is the one that
ships with a flight log rich enough to show every check). Point it at your
own log with demo /path/to/log (mount it with -v first), or keep reading
for the full MCP setup below.
⚡️ Quickstart
Two ways to run Bagel. Pick by data:
You have | Run Bagel with |
Recorded data: ROS 1 |
|
Live ROS robots (rosbridge), fleet and edge deployments | Docker, further down: the images carry the ROS stacks and the standing-pipeline runtime |
🐍 Install with uvx or pip (no Docker)
Your MCP client launches the server itself over stdio, so there is nothing to
start or keep running. Pick uvx (nothing to manage: it fetches and caches
Bagel on first use) or pip (a normal install you control).
With uvx. Install uv
(curl -LsSf https://astral.sh/uv/install.sh | sh on macOS and Linux), then:
Claude Code:
claude mcp add bagel -- uvx --from "bagel-mcp[ros]" bagel-mcp --transport stdioAny client that takes a JSON MCP config (Claude Desktop, Cursor, Codex, ...):
{
"mcpServers": {
"bagel": {
"command": "uvx",
"args": ["--from", "bagel-mcp[ros]", "bagel-mcp", "--transport", "stdio"]
}
}
}With pip. Python 3.10 to 3.12; a virtual environment keeps it tidy:
python3 -m venv ~/.bagel-venv
~/.bagel-venv/bin/pip install "bagel-mcp[ros]"
claude mcp add bagel -- ~/.bagel-venv/bin/bagel-mcp --transport stdioFor a JSON config, use the full path as the command:
"command": "/Users/you/.bagel-venv/bin/bagel-mcp", "args": ["--transport", "stdio"].
Apple Silicon Macs: use an ARM Python, not an Intel Python running through
Rosetta. Check with python3 -c "import platform; print(platform.machine())";
it should print arm64. If it prints x86_64, create a separate native environment:
uv venv --seed --python cpython-3.12-macos-aarch64-none ~/.bagel-arm-venv
~/.bagel-arm-venv/bin/python -m pip install "bagel-mcp[ros]"
claude mcp add bagel -- ~/.bagel-arm-venv/bin/bagel-mcp --transport stdioThis avoids the cryptography source build: upstream removed Intel macOS
support in version 49.
Fresh installs currently select MCP 1.30.0, whose dependency chain is
bagel-mcp → mcp → pyjwt[crypto] → cryptography. This applies to the
base package and bagel-mcp[ros], even without the upload extra.
On an Intel Mac, if installation fails while building cryptography, use the
Docker path below. Do not downgrade cryptography to restore an old wheel;
its older releases miss security fixes.
Format support comes as extras, so that a PX4 user never downloads the
automotive parsers. List the ones you need, comma-separated:
uvx --from "bagel-mcp[ros,px4,automotive]" bagel-mcp --transport stdio, or
pip install "bagel-mcp[ros,px4,automotive]". If you ask about a format whose
extra is missing, Bagel's error names the exact command to add it.
Extra | Adds |
| ROS 1 |
| PX4 |
| ArduPilot |
| Betaflight |
| CAN captures ( |
| Live MQTT (incl. Sparkplug B) and InfluxDB sources |
| Rerun export |
| GCS and Azure Blob upload tasks |
| Cloudini point-cloud tasks |
MCAP, CSV / JSON / Parquet, ROS text logs, PlotJuggler / Lichtblick / LeRobot exports and S3 upload need no extra.
PX4 log parsing does not require Git. Git and network access are used only for
optional field descriptions; pass download_description: false in the tool's
args to skip fetching those descriptions.
To upgrade, repeat the extras you installed with, so their dependencies
update too. uvx picks up new releases on its own (force it with
uvx --refresh --from "bagel-mcp[ros,px4]" bagel-mcp --help); with pip, run
pip install -U "bagel-mcp[ros,px4]", using your own list of extras.
Then prompt, pointing at your own file:
Summarize the metadata of the MCAP bag "~/logs/run_42.mcap".
No ROS installation is needed for bag files: theros extra reads and writes
them with the pure-Python rosbags library
(sqlite3 and MCAP storage, zstd / bz2 / lz4 compression, message types from the
bag itself or from the bundled interface definitions of every distro from Humble
on). Live ROS topics (rosbridge) and the fleet/edge runtime stay with Docker.
🐳 Run with Docker (live robots, fleet and edge, distro-specific ROS)
Already have Claude Code? Just paste the link to this repo and tell Claude what environment you want:
Set up https://github.com/Extelligence-ai/bagel for ROS2 Kilted.
Claude will clone the repo, start Docker, and wire up the MCP connection for you.
📋 Prerequisites
Install Docker Desktop and Claude Code (or another MCP-enabled LLM).
arm64 hosts (Apple Silicon, Raspberry Pi, Jetson, Graviton): ros2-kilted
— the default service, and the one server.json pins — ships as a multi-arch
image, so Docker pulls a native arm64 build. No extra setup.
The other services are published for amd64 only. Docker Desktop emulates
them automatically, so they run on Apple Silicon (slower, but they work). On
arm64 Linux with plain Docker Engine there is no emulation by default and
they fail immediately with exec format error — install QEMU/binfmt first:
docker run --privileged --rm tonistiigi/binfmt --install amd641. Clone and start Bagel
git clone https://github.com/Extelligence-ai/bagel.git && cd bagel
docker compose run --service-ports ros2-kiltedPort 8000 already in use? SetMCP_SERVER_PORT to something else, for example
MCP_SERVER_PORT=8100 docker compose run --service-ports ros2-kilted, and use
that port in step 2.
Pick the service that matches your environment:
Service | Use case |
| ROS2 Kilted (latest) |
| ROS2 Jazzy |
| ROS2 Jazzy + on-robot decision model (GPU, beta) |
| ROS2 Iron |
| ROS2 Humble |
| ROS1 Noetic |
| ROS1 Noetic + CV |
| PX4 flight logs |
| ArduPilot flight logs |
| Betaflight flight logs |
| IoT / MQTT (live) |
The -jev image (beta) adds PyTorch for running a decision model on the robot
(backend: local in the anomaly gate). Build any
other service the same way with --build-arg JEV_MODE=true. CPU-only robots don't need
it: the hosted Jev backend works in every image.
To give Bagel access to your local files, editcompose.yaml before starting Docker:
uncomment and update the volumes section under your chosen service.
Wait for this output:
INFO: Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)2. Connect Claude Code
In a new terminal:
claude mcp add --transport sse bagel http://localhost:8000/sseThe MCP endpoint is bound tolocalhost only (not exposed to the LAN) for security.
To share it with other machines, drop the 127.0.0.1 prefix in compose.yaml and
put an authenticated proxy in front: see SECURITY.md.
3. Prompt
claudeSummarize the metadata of the ROS2 bag "./data/sample/ros2/mcap".
That’s it: you’re chatting with your data.
🔒 Prefer fully offline?
Swap step 2 for a local model: your data and your LLM stay on the machine:
brew install ollama && ollama serve & # or ollama.com
ollama pull qwen3:8b
uvx ollmcp --mcp-server-url http://localhost:8000/sse --model qwen3:8bModel picks, expectations, and troubleshooting: Local LLMs guide.
Bagel works with any MCP-enabled LLM. Setup runbooks for tested alternatives:
Claude Code (detailed guide)
Grok Bot (runs on Grok's cloud computer)
Can’t find your LLM? Open a ticket.
🔌 Agent plugins (Claude Code, Codex and Grok Build)
Bagel ships an agent plugin: four skills that teach the agent when and how
to drive the server (log triage, pipeline authoring, live sinks, visualization
export) plus the MCP connection, wired automatically. The same plugin/
directory serves Claude Code, OpenAI Codex and xAI Grok Build.
/plugin marketplace add Extelligence-ai/bagel
/plugin install bagel@bagelCodex and ChatGPT users: install bagel from the
OpenAI Plugins Directory
(one click), or clone the repo and add it as a plugin marketplace (the repo
carries .agents/plugins/marketplace.json). Directory installs bundle the
skills only (the directory accepts only public HTTPS MCP servers, and
Bagel's runs on your machine), so also connect the server once with
codex mcp add bagel --url http://localhost:8000/mcp, or in
~/.codex/config.toml:
[mcp_servers.bagel]
url = "http://localhost:8000/mcp"Grok Build users install straight from this repo; the MCP connection is wired automatically:
grok plugin install Extelligence-ai/bagel#pluginRepo-marketplace, Claude Code and Grok Build installs wire this connection
automatically.
Maintainers build the directory ZIP with
uv run python scripts/package_codex_directory.py.
Then start the container for your data format (see Quickstart): the plugin
connects to http://localhost:8000/mcp by default. Any other MCP client can
discover the same workflows server-side via the list_agent_capabilities tool.
Keep what matters, drop the rest
A robot records more data than you can afford to move. Bagel turns a question into a detector, runs it where the data is recorded, and ships only the windows around real events.
Here it is in one conversation:
Don't know the event in advance? The anomaly gate (beta) learns what normal looks like on the robot, asks Jev to name whatever isn't, and keeps only those slices, each with a JSON label, for any bucket: S3, GCS, Azure, MinIO or R2.
The session above: a 20-minute (1,200 s) recording and the prompt "keep 10 seconds before and after every deceleration harder than −10 m/s²". The preview detects 7 events, merges them into 4 windows, and keeps 92 s of the 1,200 (7.6%); the run writes a 2.1 GB bag down to 161 MB. These figures are illustrative demo output, not a measured benchmark: the ratio is event-window duration over total duration, so it depends entirely on your workload.
✅ Supported Data Formats
Industry | Formats |
Robotics | ROS1, ROS2, MCAP (any profile), Copper (via MCAP export), ROS text logs ( |
Robot learning | Gantry Bench evidence bundles — a dataset verdict's working (per-clip signal checks, robot-test ladder, findings) as queryable tables |
Drones | PX4, ArduPilot, Betaflight |
Automotive | ASAM MDF4 ( |
IoT | MQTT (live, Sparkplug B), PostgreSQL / TimescaleDB, InfluxDB 3 |
Hardware state | WaffleForm snapshots ( |
🆚 Bagel vs. the Tools You Already Use
You already have ros2 *, PlotJuggler, and grep. Bagel doesn't replace them: it
answers the questions they make you work for, then hands off to them:
You do this today | Ask Bagel instead |
| "Summarize this bag": same prompt works on PX4, ArduPilot, MCAP, MQTT, Postgres |
| "What's the peak z-deceleration in /imu? Running average over 5 s?" · real SQL underneath: peaks, running averages, percentiles, cross-topic correlations |
Scrub PlotJuggler timelines hunting for the event | "Find every deceleration under −10 m/s² and cut ±30 s snippets": then open the result in PlotJuggler with a pre-framed layout |
| "Read the ERRORs from ~/.ros/log and tell me what went wrong": tracebacks included, no bag needed |
Echo two topics in two terminals, correlate in a spreadsheet | "What's the correlation between current and voltage?": topics live in one SQL relation, so joins and |
| A standing edge pipeline: record continuously, keep only event windows, drop the rest |
A bash loop over 200 bags | "Run this pipeline on every bag in the folder": one pipeline, whole fleet, with a combined report |
| Upload to S3, GCS, or Azure as a pipeline step, checksum-skipping files already there |
A different viewer per format: FlightPlot for PX4, MAVExplorer for ArduPilot, Blackbox Explorer for Betaflight | The same conversation for all of them, and ROS, MCAP, MQTT, Postgres, InfluxDB |
Write a one-off pandas script per question | Ask the question; Bagel writes and runs the query |
One sentence of plain language, one answer, instead of a pipeline of commands and a script you'll delete tomorrow.
💬 What Can I Prompt?
You can ask Bagel almost anything. For example:
What’s the correlation between current and voltage in the
/spot/status/battery_statestopic?
I think the robot hit a pothole. Can you check for sudden deceleration on the z-axis to confirm?
Every time the drone decelerates harder than -10 m/s², keep 10 seconds before and after. Drop everything else.
Did anything change on this robot since last week?
Time to put Bagel to the test: can it catch a drone doing barrel rolls? Spoiler: 🎉 It totally can.
💡 How Bagel Works
When you ask a question, Bagel analyzes your data source’s metadata and topics to build a high-level understanding.
Based on your prompt, if further inspection is needed, Bagel identifies the most relevant topics and interprets their meaning and structure. Bagel then writes the relevant topic messages to an Apache Arrow file and uses DuckDB to generate and execute queries against it.
This process is repeated as needed, running new queries until Bagel finds the best answer to your question.
LLMs excel at language but struggle with math. Bagel overcomes this by generating deterministic DuckDB SQL queries. These queries are displayed for you to audit, and you can guide Bagel to correct any errors.
🐶 Teach Bagel a New Trick
Bagel learns new capabilities through POML files: a structured set of instructions that describe a “trick,” such as computing latency statistics.
✍️ Create a .poml file
For example, let’s define ./bagel_mcp/agent/examples/woof.poml.
<poml>
<task>
Count the topics in the data source.
If the count is odd, say "woof", else say "meow".
</task>
<output-format>
Return the sound, the topic count, and a few cute emojis. Nothing else.
</output-format>
</poml>🗣️ Use the capability
Prompt Bagel:
Run the POML capability "./bagel_mcp/agent/examples/woof.poml" on the ROS2 bag "./data/sample/ros2/mcap".
Result:
meow 🐱 4 topics 🐱💤🎯Teach it your own tricks (no rebuild)
Bagel discovers your own capabilities from ~/.bagel/capabilities/:
In conversation: do a workflow once, then say "save that as a capability called battery-triage" — Claude calls
save_agent_capabilityand it's reusable in any future session.As a file: drop a markdown file with your steps (or a POML file, if you want parameterized templates — see
bagel_mcp/agent/compose/pipeline.pomlfor the house style) into~/.bagel/capabilities/.
Either way it shows up in list_agent_capabilities as user/<name> and runs
with run_poml_capability — from Claude Code, Claude Desktop, or any MCP
client. Teams: keep the directory in your own git repo and sync it to every
robot; it's just files. On Linux, run mkdir -p ~/.bagel/capabilities once
before starting the container so the mount is owned by you, not root.
📚 Guides
Natural-language pipelines · the model: a cadence, gates, and tasks; preview → run → save → batch → standing at the edge
Event-driven data reduction · detect events, keep windows around them (snippets or one reduced bag), batch across fleets, upload to the cloud
Anomaly detection with Jev (beta) · learn normal on the robot, ask Jev to label what isn't, upload only those slices with a JSON label;
preview_anomaliesdry-runs the screen firstLive ROS2 robots over rosbridge · a step-by-step tutorial
ROS text logs · inspect
~/.ros/logerrors and warnings without opening a bagMQTT · live IoT topics, Sparkplug B, edge recording
PostgreSQL / TimescaleDB · every table is a topic
InfluxDB 3 · every measurement is a topic
Automotive MDF4 & CAN (beta) · channel groups and DBC messages are topics; units ride along
Local LLMs · fully offline with Ollama: your data and your model never leave the machine
📦 Integrations
Rerun · "show me that event in Rerun": any time window as a ready-to-open recording
Lichtblick / Foxglove · event windows as MCAP + pre-framed layouts for either viewer
PlotJuggler · open Bagel's MCAP outputs directly; one-sentence pre-framed sessions, flattened CSV/Parquet exports
Cloudini · decode cloudini-compressed pointclouds, or compress a bag's PointCloud2 topics into CompressedPointCloud2
Slack · pipelines post to your ops channel when they fire: "🚨 hard brake on {asset}"
LeRobot (beta) · detected events become training episodes: a LeRobotDataset v3.0
🚧 Limitations
Rough edges we know about, so you don't find them the hard way:
Two formats are beta. The automotive MDF4/CAN readers are verified against files we generate with the same libraries that read them (
asammdf,python-can); real CANape/INCA/Vector-produced captures haven't crossed our test bench yet. LeRobot exports load-test clean with the reallerobotpackage, but no policy has been trained from a Bagel export yet.The Jev integration is beta, and stays beta while we learn from real deployments. That covers the
anomalyanddecidegates (recorded logs and live subscriptions),preview_anomalies, the on-robotlocalbackend and theros2-jazzy-jevimage. Its Jev backend has been run against live Jev through Vercel AI Gateway on a real drive, a synthetic fault log and a live MQTT stream; a direct TypeSafe key is not yet exercised, and detection quality has not been measured on logs with known incidents. On a live subscription the flagged window is kept as Parquet (write_topics_to_file); MCAP/rosbag snippets need a recorded log and are refused there. The baseline is learned per run and restarts with the process, so in screen mode the first minutes of each run are never flagged. Labels, settings and defaults may change between releases.Reduction ratios are workload-dependent, and unbenchmarked. The ratio is event-window duration over total duration: quiet recordings reduce dramatically, eventful ones much less. The figures in this README are illustrative demo output, not a measured benchmark.
No authentication on the MCP endpoint. By design it binds to localhost only; treat it like a database socket and see SECURITY.md before sharing it beyond your machine.
Small local models struggle with multi-step pipelines. A 4-8B model handles tool selection and simple SQL; event-windowed reduction and multi-topic joins want a bigger model. See the Local LLMs guide.
Live-database end-to-end tests run outside CI. The InfluxDB and Postgres suites' pure tests run in CI; their live end-to-end cases only execute against an instance you point them at. Everything else, including the ROS bag write paths, runs in CI.
🫶 Contributing
We’d love your help! The easiest way to support the project is by giving it a ⭐ on GitHub.
Other great ways to contribute:
Request new features
Report bugs
Improve documentation
Add new capabilities
Before contributing, please review the guidelines.
Join the conversation in our Discord server. We hang out there regularly.
📄 License
Bagel is open source under the Apache License 2.0.
Agent discovery and reproducible workflows
For maintainers: discovery audits and evaluation, listing maintenance, and reproducible user reports.
Available Tools
25 toolsdelete_capabilityDelete a user capabilityADestructiveIdempotent
Delete a capability previously saved with save_agent_capability, by the exact, full name list_agent_capabilities reports (user/-prefixed) -- a bare slug is rejected, since a user capability's name can shadow a builtin of the same stem. Only user-saved capabilities can be deleted -- builtins shipped with Bagel refuse with a clear message. An unknown name raises rather than silently no-op-ing, listing the user capabilities that do exist.
Inputs: name (required). Returns: object
delete_pipelineDelete a saved pipelineADestructiveIdempotent
Delete exactly one pipeline YAML file previously written by save_pipeline, by the same name list_pipelines reports. Confined to the trusted pipelines directory (settings.PIPELINES_DIRECTORY) -- there is no directory argument, so this can never be pointed at an arbitrary path -- and a name that would resolve outside it is refused before anything is touched. Deleting an unknown name raises rather than silently no-op-ing, listing the names that do exist -- so a second delete of the same name also raises.
Inputs: name (required). Returns: object
describe_data_sourceDescribe a data sourceARead-onlyIdempotent
Inspect a robotics, drone, or IoT log first: returns source metadata, available topics, and instructions for summarizing them. Use describe_topic next for field schemas before SQL. Does not return message rows or detect anomalies. Paths are resolved on the Bagel server; runtime and schema support depend on the selected service.
Inputs: path (required), args. Returns: result
describe_topicDescribe a topic in a data sourceARead-onlyIdempotent
Inspect one known topic before writing SQL or event predicates. Returns its DuckDB schema, original message definition, and query instructions, without message rows. Discover topic names with describe_data_source. Confirm units from the definition or user; do not infer units from field names.
Inputs: path (required), topic (required), args. Returns: result
export_for_lerobotExport event windows as a LeRobot training dataset (beta)AIdempotent
Write selected event windows as a LeRobotDataset v3.0: one episode per window, scalar signals grouped into feature vectors and resampled to a uniform fps using last observation carried forward. Returns the dataset directory, episode/frame counts, and loading instructions. Choose this for robot-learning dataset preparation, not interactive viewing or model training. Beta: load compatibility tested; actual training-run validation remains outstanding.
Inputs: fps (required), path (required), task (required), topics (required), +5 more. Returns: object
export_for_lichtblickExport an event window for Lichtblick / FoxgloveAIdempotent
Write a selected time window as JSON-encoded MCAP plus a plot layout for Lichtblick or Foxglove; returns paths, curves, and opening instructions. Choose this for those viewers, not byte-preserving native ROS/CDR export. Inspect topic schemas and use source timestamps in seconds. Automatically selects up to eight numeric plot curves unless signals are specified. Requires a separate viewer; does not launch it.
Inputs: path (required), topics (required), end_seconds (required), start_seconds (required), +3 more. Returns: object
export_for_plotjugglerExport an event window for PlotJugglerAIdempotent
Write a selected time window as flattened scalar CSV plus a PlotJuggler XML layout; returns paths, plotted curves, and an opening command. Choose this for scalar plotting in PlotJuggler, not native bag preservation. Inspect topic schemas and use inclusive source timestamps in seconds. Automatically selects up to eight numeric curves unless signals are specified. Requires the separate viewer to open; does not launch it.
Inputs: path (required), topics (required), end_seconds (required), start_seconds (required), +3 more. Returns: object
export_for_rerunExport an event window for the Rerun viewerAIdempotent
Write a selected time window as scalar time series in a Rerun .rrd recording; returns its path, signals, and an opening command. Choose this when the user requests Rerun. Requires rerun-sdk (uv sync --group viz) and a separate viewer. Inspect topic schemas and use source timestamps in seconds. This exporter does not produce camera or 3D scene replay and does not launch the viewer.
Inputs: path (required), topics (required), end_seconds (required), start_seconds (required), +3 more. Returns: object
get_pipelineRead a saved pipelineARead-onlyIdempotent
Return the full configuration of one pipeline saved by save_pipeline, by the name list_pipelines reports, so it can be inspected, edited and saved again or passed to run_pipeline. Confined to the trusted pipelines directory (settings.PIPELINES_DIRECTORY); a name resolving outside it is refused.
Inputs: name (required). Returns: object
list_agent_capabilitiesList agent capabilitiesARead-onlyIdempotent
List every capability available to run: the predefined .poml capabilities shipped with Bagel, plus any user-saved capabilities (.poml or .md, named with a user/ prefix) discovered under the user-capabilities directory. Each entry has a name, a path to pass to run_poml_capability, and a one-line summary. Use this to discover available capabilities instead of guessing file paths.
Inputs: no named parameters. Returns: result
list_live_topicsList available live topicsARead-onlyIdempotent
Discover topic names from a live broker or ROS bridge before subscribing. Supported type_ values: mqtt, ros1.bridge, ros2.bridge. Requires a reachable service; specify host and port when defaults do not fit. Returns topic names without starting a persistent recording. Use describe_data_source for an existing recorded log.
Inputs: type_ (required), args, host, port. Returns: result
list_pipeline_capabilitiesList pipeline capabilitiesARead-onlyIdempotent
Discover available task and gate modules before authoring pipeline YAML. Returns module paths, constructor parameters, summaries, and availability. These are executable building blocks, not saved workflows: use list_pipelines for saved configs and list_agent_capabilities for reusable agent instructions.
Inputs: include_unavailable. Returns: result
list_pipelinesList saved pipelinesARead-onlyIdempotent
List the pipeline YAML files saved by save_pipeline in the trusted pipelines directory (settings.PIPELINES_DIRECTORY): each entry's name, path, and a one-line summary (task count, site/asset, and cadence). Use this to discover what has already been saved before reusing, editing, or deleting it -- instead of guessing file names.
Inputs: no named parameters. Returns: result
preview_anomaliesPreview what the anomaly gate would flag (beta)ARead-onlyIdempotent
Dry-run the on-robot anomaly screen over a recorded log without calling any decision model or writing artifacts. Learns a rolling baseline the way the src.pipeline.gates.anomaly gate does, then reports every window the screen would flag (mean shift, extreme sample, topic dropout) with the signal, value and z-score, flag counts per signal, and plain-language advice (signals that drift by design, warm-up longer than the log). Inspect topics first and pass signals as rates and errors (accelerations, angular rates, currents), never positions or orientations. Use it to choose signals and thresholds before saving an anomaly pipeline. It models screen mode with the decision model confirming every flag, every: N seconds cadences, and a gate that sees every fire (list the anomaly gate first).
Inputs: path (required), window_seconds (required), args, topics, +8 more. Returns: object
preview_pipelinePreview an event-driven data reductionARead-onlyIdempotent
Preview an event-window reduction without writing artifacts. Inspect source and topic schemas first, then provide a SQL boolean predicate and nonnegative pre/post seconds. Detects false-to-true transitions, debounces nearby events, merges overlapping windows, and returns event timestamps, intervals, total seconds, and kept seconds/fraction. Report these results and obtain user confirmation before executing the reduction with run_pipeline or run_pipeline_batch. Does not predict output byte size.
Inputs: path (required), predicate (required), event_topic (required), pre_seconds (required), +3 more. Returns: object
query_messagesQuery topic messages with SQLARead-onlyIdempotent
Answer quantitative questions with read-only DuckDB SQL over one topic: filtering, aggregates, downsampling, and event evidence. Call describe_data_source and describe_topic first; use the returned schema and show the SQL to the user. Returns rows as dictionaries. Use read_loggings for textual diagnostics. Time bounds are inclusive source timestamps in seconds, not offsets from the first message.
Inputs: path (required), topic (required), sql_statement (required), args, +2 more. Returns: result
read_loggingsRead logging messages from a data sourceARead-onlyIdempotent
Read textual INFO/WARN/ERROR diagnostics from a recorded source, optionally within inclusive source-time bounds in seconds. Returns logging records, or an empty list when no logging topics exist. Use query_messages for numerical signals, statistics, and threshold detection; empty diagnostics do not prove a healthy log.
Inputs: path (required), args, end_seconds, start_seconds. Returns: result
run_pipelineRun a pipelineA
Execute one pipeline config against its single config.path after configuration and input validation. For event reductions, first call preview_pipeline, report events and kept seconds, and obtain user confirmation. Returns pipeline status, run counts, and artifact paths; inspect status for failures. Tasks may write files or contact external services. Use save_pipeline to store without executing, or run_pipeline_batch for multiple sources.
Inputs: config (required). Returns: object
run_pipeline_batchRun a pipeline across many data sources (batch)A
Execute one pipeline config for multiple explicit source paths or globs, overriding config.path for each source. Each source runs independently; returns per-source statuses/errors and totals. For event reductions, preview each source to be reduced and obtain confirmation of the batch scope before executing. Tasks may write files or contact external services. Use run_pipeline for a single source. A glob matching nothing is treated as a literal path.
Inputs: paths (required), config (required). Returns: object
run_poml_capabilityRun a capability defined in a POML fileARead-onlyIdempotent
Load reusable agent instructions from a POML or Markdown file; returns a prompt for the calling agent to follow. Discover paths with list_agent_capabilities. POML accepts template context; Markdown rejects nonempty context. Loading the prompt does not itself analyze data, execute a pipeline, or write reduction artifacts. Use run_pipeline for an approved executable pipeline config.
Inputs: poml_path (required), poml_context. Returns: result
save_agent_capabilitySave a user capabilityADestructiveIdempotent
Save a reusable workflow as a named capability so it can be discovered with list_agent_capabilities and run with run_poml_capability in any future session. Content may be POML (validated before saving; supports context parameterization) or plain markdown instructions. Writes only to the user-capabilities directory; builtin capabilities cannot be modified.
Inputs: name (required), content (required), overwrite. Returns: object
save_pipelineSave a pipeline to a YAML fileBIdempotent
Persist a pipeline configuration to a YAML file so it can be reused, edited, or run later with run.py. Returns the path to the written file.
Inputs: name (required), config (required), directory. Returns: result
snap_hardwareSnapshot robot hardware into a WaffleForm (experimental beta)A
Auto-detect the robot's current hardware, firmware, and software using waffle-iron and return the resulting hardware state. Requires the waffle CLI on PATH (cargo install waffle-iron). The WaffleForm it writes is immediately queryable as a data source.
Inputs: directory. Returns: object
subscribe_live_topicsSubscribe to live topic messagesADestructive
Start a background subscription to mqtt, ros1.bridge, or ros2.bridge and persist messages locally; returns the sink directory for subsequent analysis. Use list_live_topics first. An optional pipeline runs continuously on incoming messages and may write or upload artifacts. overwrite=True clears existing topic buffers. Stop it with unsubscribe_live_topics. Not for inspecting an existing file.
Inputs: type_ (required), args, host, port, +3 more. Returns: result
unsubscribe_live_topicsStop live topic subscriptionsAIdempotent
Stop topics started by subscribe_live_topics on a running mqtt, ros1.bridge, or ros2.bridge subscription (same type_, host and port). Omit topics to stop all of them. A standing pipeline finishes its queued runs and its end-of-stream run first, which may write or upload artifacts. Recorded messages stay in the returned sink directory for analysis. Once nothing is subscribed the connection is closed. Never opens a connection: errors if there is no running subscription there.
Inputs: type_ (required), host, port, topics. Returns: object
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 25 tools
Each tool targets a distinct resource or action: capabilities, pipelines, data-source inspection, live-topic subscriptions, and exports are all clearly separated. Similar-sounding tools such as run_poml_capability versus run_pipeline, list_pipelines versus list_pipeline_capabilities, and query_messages versus read_loggings are explicitly cross-referenced and disambiguated in their descriptions.
The set overwhelmingly follows a verb_noun snake_case pattern with consistent list/save/delete/run/preview/export verbs. Minor deviations include run_poml_capability supporting Markdown despite the POML-specific name, delete_capability breaking the save_agent_capability/list_agent_capabilities symmetry, and the somewhat vague snap_hardware verb.
With 25 tools the server sits exactly in the heavy 16-25 range, making it borderline even though the domain is fairly broad. The pipeline and capability management clusters plus the four exporter tools add significant surface area, though none of the tools feel entirely redundant.
Core lifecycles are covered well: capabilities have list/save/run/delete, pipelines have save/get/list/delete/run/batch/preview, and data inspection covers schemas, SQL queries, and textual logs. Minor gaps remain, such as no tool to enumerate recorded data sources or produce byte-preserving native bag exports, but agents can work around these.
Maintenance
Related MCP Connectors
Query your warehouse or a CSV with Claude/ChatGPT over MCP, governed by table-level ACL + audit.
Analytical memory for AI agents: a real Postgres queried in plain English over MCP. One command.
Query, join, profile, clean and convert CSV/JSON/Parquet with server-side DuckDB over MCP.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables language models to perform hardware engineering tasks including CAD part design and heat transfer simulations. Provides tool calls for building mechanical components and running thermal analysis through natural language interactions.-
- AlicenseCqualityBmaintenanceProvides unified control for both physical robots (ROS-based like Moorebot Scout, Unitree) and virtual robots in Unity3D/VRChat, enabling multi-robot coordination, environment generation, and automated 3D model creation.2512MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to control hardware devices like Arduino, Raspberry Pi, 3D printers, CNC machines, and custom robots via serial ports and HTTP. Provides tools for device discovery, command sending, sensor reading, servo control, G-code execution, and emergency stops with safety features.37 PyPIMIT
- AlicenseAqualityDmaintenanceProvides an MCP interface to the ROBOT command-line tool for OWL ontology editing, enabling operations like merging, reasoning, and conversion via natural language.2636 PyPI1BSD 2-Clause "Simplified"