01Safety
0.0 / 30
What changed in the harness
Selection accuracy 100→100, token cost up 14%, 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
29.5 / 30
03Economics
20.0 / 20
04Discoverability
18.6 / 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 |
|---|---|---|
run_process |
name_restates_behavior no_return_description |
Execute a process on this darwin machine in either shell mode (a command line run via the system's default shell, supporting pipes, redirects, and globbing) or executable mode (spawned directly with argv[0] as the executable and arguments passed verbatim without shell interpretation). These two modes are mutually exclusive. Optionally set the working directory with cwd, write text to STDIN with stdin_text (written fully, then closed), and control the wait with timeout_ms (defaults to 30,000ms). Returns the result of the process execution, including its output and exit status. |
Selection evidence
Confusable tool pairs.
0 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
| No confusable tool pairs were flagged in this assessment. | |||
Compare the field