Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →mkurman/parallel-web
Search the web, extract URL content, and run deep research using the Parallel Chat API and Extract API.
Passed 23 of 23 test cases with the skill, and 4 of 23 without it.
and 27 more in the full report
19 fixed4 passed either way0 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 19 test cases the skill fixed. Same model, same prompt, with and without the skill.
Write the CLI command for a deep research query on global critical mineral recycling policies on 2025-05-20 at 14:00:00, ensuring the file is saved according to the project's standard naming structure.
| Checked | Without skill | With skill |
|---|---|---|
| The command executes research | Passed | Passed |
| The output file argument follows sources/research_20250520_140000_ | Missed | Passed |
The instructions your model reads when the skill loads.
The Chat API supports two research models. Use base for most searches and core for deep research.
| Model | Latency | Strengths | Use When |
|---|---|---|---|
base | 15s-100s | Standard research, factual queries | Web searches, quick lookups |
core | 60s-5min | Complex research, multi-source synthesis | Deep research, comprehensive reports |
Recommendations:
search command defaults to base — fast, good for most queriesresearch command defaults to core — thorough, good for comprehensive reports--model when you need different depth/speed tradeoffsfrom parallel_web import ParallelSearch
searcher = ParallelSearch()
result = searcher.search(
objective="Find latest information about transformer architectures in NLP",
model="base",
)
if result["success"]:
print(result["response"]) # Synthesized summary
for src in result["sources"]:
print(f" {src['title']}: {src['url']}")
from parallel_web import ParallelDeepResearch
test cases passed, vs 4 / 23 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.