Alchemist-MCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| get_pipeline_statusA | Return lifecycle and gate status without expensive execution. |
| inspect_dft_datasetB | Inspect QE outputs without mutating accepted data. |
| verify_dft_datasetC | Run available fail-closed verification over incoming QE outputs. |
| freeze_verified_datasetC | Freeze only after Gate 1-3 eligibility is proven. |
| run_mace_pipelineC | Return a fail-closed MACE dry-run plan; never launches MACE. |
| run_runner2_pipelineC | Return a fail-closed Al-only RuNNer 2G dry-run plan; never launches RuNNer. |
| compare_modelsC | Compare model arrays only when ordered common-test record IDs match. |
| validate_referenceC | Validate barriers against supplied DFT/reference evidence without literature claims. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 8 tools
Tools target distinct operations (status, inspect, verify, freeze, run dry-runs, compare, validate). However, run_mace_pipeline and run_runner2_pipeline are very similar in nature (dry-run plans for different models), and inspect_dft_dataset and verify_dft_dataset could be confused as both deal with QE outputs. Descriptions clarify the differences, so ambiguity is minimal.
All tool names follow a consistent verb_noun pattern (get_pipeline_status, inspect_dft_dataset, verify_dft_dataset, freeze_verified_dataset, run_mace_pipeline, run_runner2_pipeline, compare_models, validate_reference). Even multi-word nouns are handled uniformly, making the naming highly predictable.
Eight tools provide a well-scoped surface for a specialized DFT/MACE/RuNNer workflow. Each tool corresponds to a distinct step or check in the pipeline, and no tool appears redundant or excessive.
The tools cover inspection, verification, freezing, dry-run planning, comparison, and reference validation, but there are notable gaps: no tool to actually execute a pipeline (only dry-runs), no tool to retrieve or manage datasets beyond inspection, and no tool to handle pipeline failures or updates. These gaps could hinder agents from performing end-to-end workflows.