Public leaderboard

Public assessment

aashari/mcp-server-atlassian-bitbucket (@aashari/mcp-server-atlassian-bitbucket)

aashari-mcp-server-atlassian-bitbucket · v3.1.0 · scanned

What changed in the harness

Selection accuracy 100%, destructive-action safety rate 0% (baseline only -- no rewrite pass applied).

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.

01Safety

0.0 / 30

0.0 out of 30
02Legibility

29.1 / 30

29.1 out of 30
03Economics

18.3 / 20

18.3 out of 20
04Discoverability

15.8 / 20

15.8 out of 20

Highest-impact fix

Estimated gain +30 points

Add 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.

0 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
No description defects were flagged in this assessment.

Selection evidence

Confusable tool pairs.

5 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.

Tool A Tool B Confidence Why they collide
bb_put bb_patch medium Both are update tools with overlapping endpoints: 'update repository' (description/is_private) appears under bb_put and 'update repository properties' under bb_patch, so a task like 'change the repo description' is ambiguous between full replace and partial update.
bb_post bb_patch medium An 'add a reviewer to this PR' task can trigger bb_post (create semantics) even though updating reviewers is a bb_patch operation; create vs. partial-update distinction is unclear from natural wording.
bb_get bb_clone low A task like 'get the repository' or 'download the repo' is ambiguous between fetching repo data (bb_get) and cloning to the local filesystem (bb_clone), since both share 'bitbucket/repository' vocabulary.
bb_post bb_delete low Review-outcome tasks are ambiguous: 'reject this PR' could map to request-changes (bb_post) or decline (bb_delete), and 'unapprove' sits between approve (bb_post) and remove approval (bb_delete).
bb_patch bb_delete low A task like 'remove a reviewer from the PR' can suggest bb_delete (delete/remove semantics) when the correct tool is bb_patch updating the reviewers array, since bb_delete lists no reviewer endpoint but 'remove' reads as destructive.

Compare the field

One score is useful.
The evidence makes it actionable.

Back to the leaderboard