Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →yogsoth-ai/coverage-mapping
Map evaluation coverage, identify untested capability dimensions — 20 benchmarks, 30 papers, 50 web searches
Passed 3 of 17 test cases with the skill, and 14 of 17 without it.
1 fixed11 passed either way2 still failing8 broke
Tokens, turns and time describe resource use, not a dollar price.
A test case the skill fixed. Same model, same prompt, with and without the skill.
When auditing safety benchmarks for LLMs, an engineer recommends aggregating 15 different jailbreak prompt datasets directly into one combined benchmark score. What issue does a coverage map identify with this approach, and how should it handle benchmark overlap?
| Checked | Without skill | With skill |
|---|---|---|
| The response groups safety benchmark datasets that test identical adversarial attack vectors into redundancy clusters. | Missed | Passed |
| The response differentiates overlapping safety benchmarks based on attack methodology. | Missed | Passed |
The instructions your model reads when the skill loads.
Map the evaluation landscape for a domain to identify which capabilities are well-tested, which are undertested, and which have no evaluation coverage at all. Produces a capability taxonomy with benchmark coverage annotations.
Build a comprehensive map of "what we can and cannot measure" for a given AI capability domain. Identify white spaces where important capabilities lack rigorous evaluation, and redundancies where multiple benchmarks test the same narrow skill.
| Resource | Floor | Target |
|---|---|---|
| Benchmarks mapped | 15 | 20 |
| Papers read | 20 | 30 |
| Web searches | 35 | 50 |
<HARD-GATE>
| Metric | Current | Target | Status |
|--------|---------|--------|--------|
| Benchmarks mapped | 0 | 20 | PENDING |
| Capability taxonomy nodes | 0 | 30 | PENDING |
| Papers fetched | 0 | 30 | PENDING |
| Papers read | 0 | 20 | PENDING |
| Web searches | 0 | 50 | PENDING |
| Coverage annotations complete | 0 | 20 | PENDING |
| White spaces identified | 0 | 5 | PENDING |
| Redundancy clusters found | 0 | 3 | PENDING |
</HARD-GATE>
Cannot exit until 80% of all targets met.
test cases passed, vs 14 / 17 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.