01Safety
0.0 / 30
What changed in the harness
Selection accuracy 100→100, token cost down 0%, 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
29.0 / 30
03Economics
18.1 / 20
04Discoverability
12.4 / 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 |
|---|---|---|
fizzy_comment |
params_unexplained |
Manage comments on a card: create, list, update, or delete. **Actions:** - `create` (default): Post a new comment - `list`: Get all comments on a card - `update`: Edit an existing comment (requires `comment_id` + `body`) - `delete`: Remove a comment (requires `comment_id`) **Arguments:** - `action` (optional): "create" | "list" | "update" | "delete" (default: "create") - `account_slug` (optional): Uses session default if omitted - `card_number` (required): The card's human-readable `#` number from URLs/lists (e.g., 42) that the comment targets - `comment_id` (optional): Comment UUID. Required for update/delete - `body` (optional): Comment body in markdown. Required for create/update **Returns:** JSON with comment details, list of comments, or deletion confirmation. |
Selection evidence
Confusable tool pairs.
4 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
fizzy_comment |
fizzy_step |
medium | Both operate on a specific card via card_number and share create/update/delete semantics. A task like 'add a note about X to card 42' or 'put the requirements on the card' is ambiguous between posting a comment and adding a checklist step, since neither description strongly disambiguates 'note' vs 'checklist item'. |
fizzy_boards |
fizzy_search |
medium | Both are read/overview tools scoped to an account: boards returns boards with column card counts, search finds cards within a board. A natural task like 'show me what's on my board' or 'review the contents of the board' could plausibly route to either tool based only on the descriptions. |
fizzy_search |
fizzy_comment |
low | A task phrased as 'find/search for the comment mentioning X on a card' could lead an agent to fizzy_search (text search over cards) instead of fizzy_comment's list action, since both appear to surface card-related content, though the comment tool clearly requires a known card_number. |
fizzy_account |
fizzy_boards |
low | Both are account-scoped 'list' operations, and a vague task like 'what do I have access to' or 'list what's in my account' could route to either the account list tool or the boards-list tool; the shared account/list token reinforces the ambiguity. |
Compare the field