Generate and inspect portable Vensim .mdl models. Static validation is not a replacement for Vensim's Model Check or Units Check. On macOS, PLE can be opened for authoritative interactive validation; unattended command files require Vensim DSS.
Vensim MCP has 7 tools with basic descriptions and minimal parameter documentation. All tools have descriptions (10-100 chars), but most are terse and lack actionable context for LLM selection. Input schemas are present but sparse: most parameters lack type constraints, enums, or format guidance. No output schemas are documented. Error handling is absent, tools return raw dict responses with no recovery guidance. Tool names follow verb_noun convention (good), but descriptions do not explain WHEN to use each tool or what prerequisites exist. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk profiles (WRITE, IRREVERSIBLE operations). Parameter descriptions are minimal or missing entirely.
Create a Vensim DSS .cmd batch script; PLE does not support this official path.
Generate a styled, runnable .mdl file from a JSON model specification.
Read equations, sketch objects, arrows, and views from a text .mdl model.
Open a .mdl in the locally installed Vensim application.
Render an SVG preview from the model's actual Vensim sketch records.
Run a conservative static audit before opening the model in Vensim.
Detect local Vensim and report the exact native automation boundary.
No output schemas documented. Tools return raw dicts (e.g. {'model': str(output), 'validation': ...}) without specifying field types, structure, or what downstream tools can expect. LLMs cannot plan chained calls or extract required IDs.
Parameter descriptions are missing or trivial. 'spec_json' has description 'JSON model specification' (28 chars), does not explain format, required fields, or example structure. 'output_path' lacks any guidance on absolute vs relative paths, directory creation, or overwrite behavior.
No tool annotations despite clear risk profiles. 'open_vensim_model' is IRREVERSIBLE (launches external app), 'generate_vensim_model' and 'render_vensim_preview' are WRITE operations, but no destructiveHint or readOnlyHint annotations present. Agents cannot reason about safety.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
No error handling or recovery guidance. Tools return raw dicts with no error classification, actionable messages, or next steps. If 'spec_json' is malformed, the LLM receives a stack trace, not 'Invalid JSON: expected object with keys [variables, equations, views]'.
Parameter types lack constraints. 'run_name' in create_vensim_dss_script has no length limit, character restrictions, or enum. 'output_path' in multiple tools has no validation for directory existence, write permissions, or file extension requirements.