01Safety
0.0 / 30
What changed in the harness
Selection accuracy 83→92, token cost up 5%, 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
26.2 / 30
03Economics
20.0 / 20
04Discoverability
10.7 / 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.
1 defect 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 |
|---|---|---|
generate_report |
name_restates_behavior no_return_description |
Generate the final HTML market intelligence report containing all currently ranked opportunities. Call market_query or get_discover_data first to populate the opportunity data, then use this to compile the full report. Returns a complete, self-contained HTML document ready to display directly, optionally scoped by the focus category. |
Selection evidence
Confusable tool pairs.
1 pair where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
render_report |
generate_report |
medium | Both tools produce a 'full HTML ... report', so a task like 'render the report' or 'generate the full report' looks satisfied by either. The two share the token 'report' plus overlapping phrasing, and the verb difference (render vs generate) is a weak disambiguator — a user asking to 'generate the discovery report' or 'render the market report' could plausibly land on the wrong one. |
Compare the field