| FMRS | 50 / 100 · D | 79 / 100 · B |
|---|
| Reliability | 7 / 20 | 12 / 20 |
|---|
| Security and permissions | 8 / 20 | 16 / 20 |
|---|
| Maintenance | 8 / 20 | 18 / 20 |
|---|
| Documentation | 14 / 20 | 18 / 20 |
|---|
| Setup experience | 13 / 20 | 15 / 20 |
|---|
| Best for | - Developers who run LLMs locally
- Users needing hardware-aware model selection and Ollama management
- Teams requiring deterministic scoring, calibrated routing, or policy audits
- Users who want structural checks before running downloaded models
| - Developers auditing NPM dependencies within AI workflows
- Teams performing supply chain security assessments
- Users of Claude Desktop, Cursor, or VS Code
|
|---|
| Not for | - Users without Node.js or local Ollama
- Cloud-hosted inference workflows rather than local inference
- Complete guarantees about model behavior, provenance, or poisoned weights
- Arbitrary system command execution beyond the MCP allowlist
| - Projects outside the NPM ecosystem (e.g., pure Python/Go)
- Environments without network access to deps.dev, OSV.dev, and the npm registry
- Scenarios requiring maintenance by an official upstream vendor
|
|---|
| Required permissions | - Read local CPU, GPU, memory, and acceleration information
- Access local Ollama model directories and the running Ollama environment
- Depending on the tool, download, run, or remove models and execute prompts
- Read specified project directories or model files for project recommendations and verification
- Execute allowlisted LLM Checker CLI commands
| - Network access to the NPM registry and external services (deps.dev, OSV.dev, OpenSSF, npms.io, GitHub)
- Read access to workspace lockfiles (pnpm-lock.yaml, package-lock., yarn.lock) for cache invalidation
|
|---|
| Risks and side effects | - ollama_pull downloads models and consumes network bandwidth and disk space.
- ollama_remove deletes local models and the data they occupy.
- ollama_run and cli_exec can run local model or CLI workflows, so prompts and arguments should be reviewed.
- Structural acceptance only indicates that a file is structurally safe to load; it does not establish model behavior, provenance, or absence of poisoned weights.
- Files above 3 GiB may exceed the WASM32 verification memory ceiling and produce no verdict.
| - Third-party READMEs and changelogs are untrusted external content; they are wrapped in tags with _meta flags but should still be handled cautiously
- Depends on availability and accuracy of external services
- Batch requests are capped at 25 packages to prevent registry enumeration; larger sets require batching
- Third-party open source project, not officially maintained by NPM or Anthropic
|
|---|
| Supported clients | Claude Code, Codex, Grok, Kimi Code, Cursor, Windsurf, Gemini CLI | Claude Desktop, VS Code, Cursor |
|---|
| Tools | 28 | 19 |
|---|