This MCP server has fundamental definition quality issues that prevent confident production deployment. Tool schemas are visible and mostly complete, but descriptions lack depth and LLM-optimization. Critical gaps: (1) Most tool descriptions are under 100 chars and lack guidance on when/why to use them. (2) No output schemas documented for any tool, LLMs cannot predict return structure. (3) Parameter descriptions are sparse or missing, e.g., 'process_id' in check_calculation_status has no guidance on format, and 'script_path' in execute_calculation lacks validation hints. (4) No error handling guidance, what happens if a calculation fails? How should the LLM retry? (5) Tool composition issues: generate_phonon_script and execute_calculation could be merged into a single 'run_phonon_calculation' tool since they're nearly always called together. (6) No idempotency or dry-run support for destructive operations (execute_calculation, install_code). Server lands in the 'poor/D' range due to missing output documentation and minimal parameter descriptions.
Check status of running calculations
Check if AiiDA codes and pseudopotentials are installed
Execute the phonon calculation script
Generate Python script for phonon calculation
Install an AiiDA code from YAML configuration
Install pseudopotential library
No output schemas documented for any tool. LLMs cannot predict response structure, forcing blind composition of multi-step workflows.
execute_calculation lacks error guidance and irreversibility warning. Description does not state that this operation runs expensive HPC simulations with unrecoverable side effects.
Parameter descriptions sparse or missing context. 'script_path' in execute_calculation has no format hints, validation rules, or guidance on what to do if the file doesn't exist. 'process_id' in check_calculation_status lacks format specification.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 46 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 22 | - | v1 |
Tool descriptions under 100 chars lack LLM-optimization. Most fail to answer: What does it do? When to use it vs similar tools? What does it return? Example: 'Install an AiiDA code from YAML configuration' (52 chars) omits when to call it (after check_prerequisites?) or what happens on success.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Agents cannot determine which tools are safe to retry without failing destructively. execute_calculation is marked IRREVERSIBLE in risk but has no hint in the tool definition.
Composition issue: generate_phonon_script and execute_calculation are tightly coupled and nearly always invoked together. Consider merging into 'run_phonon_calculation' or adding a confirmation step before irreversible execution.
No dry-run or confirmation capability for destructive operations. install_code, install_pseudopotentials, and execute_calculation modify system state with no rollback guidance or confirmation prompt.