list_projectsLABSCO SUMMARY
Deletes have no undo.
The tools that matter most in daily use are the boring ones — batch upload a folder, check per-document progress, export finished annotations in the format your parser wants. The one thing to be careful about is stated plainly in the README: delete_project, delete_document, delete_annotations and delete_curation are irreversible in INCEpTION, so verify the IDs before an agent calls them, and prefer a project ZIP export as a backup first.
What it is
A wrapper around INCEpTION's AERO v1 REST API. It exposes corpus management as MCP tools and as a CLI: projects, documents, annotations and curation. Annotation itself stays with human annotators in the INCEpTION interface — this automates everything around it.
What you get
- Project management — list, create, delete, full ZIP export and import, and per-document progress —
list_projects,create_project,delete_project,export_project_zip,import_project_zip,project_status - Document handling including bulk work —
list_documents,upload_document,batch_uploadover a folder with a glob,export_document_source,delete_document - Annotations per user: list, export one document or a whole project, import, delete —
list_annotations,export_annotations,export_all_annotations,import_annotations,delete_annotations - The curation layer separately —
export_curation,delete_curation - 13 export formats, including
ctsv3,xmi,conllu,conll2003,conll2012,textandjsoncas - The same operations as terminal commands —
inception-cli batch-upload --project 1 --folder ./texts --glob "*.txt"— for pipelines with no agent in them
Requirements
Python ≥ 3.10 and a running INCEpTION instance with the AERO v1 REST API enabled; the README notes it is tested with INCEpTION 33+. The API lives at http://<YOUR_HOST>:<YOUR_PORT>/api/aero/v1 and authenticates with HTTP Basic using your INCEpTION credentials, supplied as INCEPTION_URL, INCEPTION_USER and INCEPTION_PASSWORD in the environment or a .env. Install with uv sync or pip install -e . from a clone; the documented client entry runs uv run --directory <YOUR_PATH_TO_INCEPTION_MCP> inception-mcp.
Setup effort
Build from source — git clone https://github.com/RaphFaure/inception-mcp
USE CASES
01
Loading a corpus into a project
Upload a whole folder in one call instead of adding documents through the interface one at a time.
02
Checking how far annotation has got
Read per-document progress per user rather than clicking through the monitoring page.
03
Getting annotations into a parsing script
Export the whole project for one annotator in the format your pipeline reads.
04
Backing up before a risky change
Export the full project ZIP — schema, documents and annotations — before deleting anything.
05
Moving a project between instances
Export the ZIP from one instance and import it into another.
18 tools
Listed in project docs. Your client may show a different list.
create_projectdelete_projectexport_project_zipimport_project_zipproject_statuslist_documentsupload_documentbatch_uploadexport_document_sourcedelete_documentlist_annotationsexport_annotationsexport_all_annotationsimport_annotationsdelete_annotationsexport_curationdelete_curationBEFORE YOU INSTALL
What to know first
Tools you get
18 tools named in the project's documentation.
Requires
- Python ≥ 3.10
- INCEPTION_PASSWORD in your environment
Where it runs
Both documented clients run the server on your machine.
INSTALLATION
Connect your client
You clone this one yourself.
Server setup
TERMINAL
git clone https://github.com/RaphFaure/inception-mcp
cd inception-mcp
uv syncMCP Clients
Claude Desktop, Claude Code
JSON
{
"mcpServers": {
"inception": {
"command": "uv",
"args": [
"run",
"--directory",
"<YOUR_PATH_TO_INCEPTION_MCP>",
"inception-mcp"
],
"env": {
"INCEPTION_URL": "http://localhost:8080",
"INCEPTION_USER": "admin",
"INCEPTION_PASSWORD": "<YOUR_INCEPTION_PASSWORD>"
}
}
}
}
