01Safety
30.0 / 30
What changed in the harness
Selection accuracy 95→98, token cost up 12%, unconfirmed writes 100%→100%.
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
26.8 / 30
03Economics
20.0 / 20
04Discoverability
15.1 / 20
Highest-impact fix
Estimated gain +5 pointsMake target tools discoverable on the first call
Clarify tool names, decision boundaries, and required argument schemas so an agent can choose and construct the target call without exploratory steps.
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 |
|---|---|---|
get_clojars_latest_version |
name_restates_behavior no_return_description |
Queries Clojars for the most recently released version of the given Maven artifact, useful for comparing against the version pinned in a project. Returns the latest published version string for the dependency. |
check_clojars_version_exists |
name_restates_behavior no_return_description |
Validates whether a specific version of a Clojars dependency is published, typically to confirm a version is safe to reference before using it. Returns a boolean indicating whether that exact version exists. |
get_clojars_history |
name_restates_behavior no_return_description |
Lists the published versions of a Clojars dependency in the order they were released, useful for range checks or finding what is available to upgrade to. Returns a list of version strings, with the number of entries controlled by the limit parameter (default 15, max 100). |
Selection evidence
Confusable tool pairs.
3 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
get_clojars_latest_version |
get_clojars_history |
medium | A task like 'what's the most recent release of metosin/reitit?' or 'show me the newest published versions' maps clearly to latest, but an agent told to 'get the latest and tell me recent versions' may gravitate to history and derive the newest, or pick latest for what is really a history request, since both share version/dependency vocabulary. |
get_clojars_latest_version |
check_clojars_version_exists |
medium | A task such as 'is 1.0.2 the current version?' or 'check if the latest version is what my project pins' could trigger the existence-check tool when the agent actually needs the latest version, or the latest tool when the user only wants a yes/no existence check; both descriptions mention version and dependency with no cross-reference. |
check_clojars_version_exists |
get_clojars_history |
medium | A task like 'was version 0.7.2 ever released?' or 'find out if a specific version appears in the published versions' can be answered via check or by pulling history and scanning; the agent may pick the list tool to see everything or the check tool for a targeted search, since neither description signals which is preferred for membership-style queries. |
Compare the field