The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Edgedelta MCP Server listing page.
The Edge Delta MCP Server is a Model Context Protocol (MCP) server that provides seamless integration with Edge Delta APIs, enabling advanced automation and interaction capabilities for developers and tools.
The server exposes tools to inspect and change Edge Delta pipelines. Write operations follow a safe loop: validate → save → deploy, with rollback.
| Tool | What it does |
|---|---|
get_pipelines / get_pipeline_config / get_pipeline_history | Discover pipelines and read their config and version history. |
validate_pipeline_config | Validate a config YAML without saving it. Returns {valid, reason}. |
save_pipeline_config | Save the full config YAML as a new version (does not deploy). |
deploy_pipeline | Deploy a saved version to the fleet. |
revert_pipeline | Roll back to a previously saved version. |
add_pipeline_source | Add a data source node to a pipeline. |
Typical edit flow: get_pipeline_config → edit YAML → validate_pipeline_config
→ save_pipeline_config → get_pipeline_history → deploy_pipeline
(revert_pipeline to roll back).
Validating that a config is valid is not the same as confirming a change actually transformed the data. The live-capture tools sample real before/after data flowing through pipeline nodes on running agents, so an agent can verify a transformation took effect.
| Tool | What it does |
|---|---|
start_pipeline_capture | Start a capture task on a pipeline (duration, optional node list, max items). |
get_pipeline_capture_status | Poll per-agent status for a capture task. |
get_pipeline_capture_results | Fetch captured before/after samples per node. |
get_pipeline_capture_task | Show the currently active capture task, if any. |
Capture flow: start_pipeline_capture → get_pipeline_capture_status (poll until
an agent reports success) → get_pipeline_capture_results (compare each node's
before vs after to confirm the transformation).
Develop and verify a processor against sample logs offline — no save, no deploy, no live traffic. The offline complement to live-capture: capture proves what did happen, this proves what will happen.
| Tool | What it does |
|---|---|
test_pipeline_ottl | Dry-run OTTL statements against sample OTEL log items; returns the transformed items. |
test_pipeline_node | Dry-run a full processor node of any type (log_to_metric, log_to_pattern, extract_metric, parse_json_attributes, grok, mask, …); returns the items it emits per output path. |
Iterate until the output is right, then add the statements/node to the pipeline,
save_pipeline_config → deploy_pipeline, and confirm on live data with the
capture tools. Use test_pipeline_ottl for OTTL transforms (its ExtractPatterns
also covers regex field extraction); use test_pipeline_node for non-OTTL node
types such as log-to-metric and log-to-pattern.
First‑time setup (creates a multi‑platform builder and boots it):
Build the image and load it into the local Docker daemon:
ℹ️ The
--loadflag streams the image back to your local Docker engine so you can run it directly withdocker run mcp/edgedelta ….
The exported Go API of this module is experimental and may change without notice. If you rely on it in production, please open an issue describing your use case so we can stabilise the relevant surface.
Licensed under the terms of the MIT licence. See LICENSE for full details.