01Safety
0.0 / 30
What changed in the harness
Selection accuracy 100→100, token cost up 3%, unconfirmed writes 0%→0%.
Category breakdown
Where the score comes from.
Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.
02Legibility
28.5 / 30
03Economics
20.0 / 20
04Discoverability
10.2 / 20
Highest-impact fix
Estimated gain +30 pointsAdd explicit identity and permission preflight tools
Expose machine-readable principal/tenant confirmation and a non-mutating permission check so agents can verify both before destructive actions.
Description evidence
Defects and rewrites.
3 defects found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.
| Tool | Defect types | Suggested rewrite |
|---|---|---|
search_mcp_servers |
no_return_description |
Keyword + semantic search across every MCP server mcpindex indexes. Returns a ranked list of matching servers, including each server's slug, name, category, and quality score. Use when the user knows what tool category they want but not which server. |
compare_servers |
no_return_description |
Side-by-side comparison of 2-5 MCP servers by slugs. Returns structured per-server data covering quality scores, install paths, and required env vars, so the user can choose between candidate servers. |
assess_server |
no_return_description |
Aggregated pre-flight trust assessment across all tools on an MCP server. Returns an advisory verdict object with the same shape as check_tool_trust (directive, dimensions, freshness), aggregated across the server's tools. Use for "is THIS server worth integrating?" decisions. v1 advisory; conformance monitored not enforced; verdicts may be UNVERIFIED if not yet probed. |
Selection evidence
Confusable tool pairs.
2 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
compare_servers |
assess_server |
medium | Both take server slugs and evaluate servers, and users often use 'assess' and 'compare' as synonyms. A request like 'assess these two servers and tell me which to trust' could route to compare_servers (quality scores, install paths) when the user actually wants the trust verdict from assess_server, or vice versa. |
recommend_mcp_for_task |
search_mcp_servers |
high | Both are discovery/selection tools for 'find the right MCP server for X'. Recommend is prompt-engineered for natural-language tasks, while search is keyword/category-based, so a user asking 'which server should I use for reading PDFs and writing to S3?' could plausibly be sent to either tool and still look superficially correct. |
Compare the field