how do i configure source policy to limit parallel search to approved domains? use include_domains
how do i configure source policy to limit parallel search to approved domains? Set source_policy with include_domains to your approved apex domains: on the Parallel Search API it nests under advanced_settings, and only sources on the list are searched. Setting include_domains makes exclude_domains be ignored, so pick one list per request. Apex domains cover their subdomains; do not use wildcard syntax.
TL;DR
To limit Parallel search to approved domains, set sourcepolicy with includedomains to the apex domains you trust: on the Search API the policy nests under advancedsettings, and only sources matching the list are searched, nothing else. An apex entry like example.com covers its subdomains automatically. If you set includedomains, exclude_domains is ignored, so pick one list per request.
how do i configure source policy to limit parallel search to approved domains?Use this when
- You need Parallel search results only from approved or official domains (compliance-bound corpora, single-publisher research, official docs only)
- Forum threads or content farms are polluting results and you want a hard domain filter, not a soft preference
- You are configuring source policy on the Parallel Search, Task, Monitor, or Responses API
Not for this skill when
- You only have a soft preference (prefer official docs over blogs). Steer that in the objective text instead of a hard filter; an include list drops relevant pages that live outside it and result quality falls sharply
- You want to combine an allowlist and a denylist in one request. That does not work: the include list wins and the exclude list is ignored
Steps
- Build the approved domain list in apex form. Write each entry as example.com with no scheme, no port, and no www. prefix. An apex domain automatically covers its subdomains. A bare extension with a leading period (.gov, .org, .co.uk) matches that whole extension. Do not use wildcard syntax. You can list up to 200 entries, though a short targeted list returns better results.
Expected: a short list of clean apex domains, ready to paste into the request.
- Add sourcepolicy with includedomains to your request. On the Search API, nest it under advanced_settings, like this:
{
"objective": "recent open source LLM benchmark reports",
"search_queries": ["open source LLM benchmarks 2026"],
"advanced_settings": {
"source_policy": {
"include_domains": ["example.com", "example.org"]
}
}
}Expected: the request is accepted with the policy in place; only sources on the list can appear in results.
- Or block instead of allow, with excludedomains. If the goal is just to keep noisy domains out (for example reddit.com when forum threads pollute results), use sourcepolicy with excludedomains instead, and do not set includedomains in the same request.
Expected: results contain no sources from the excluded domains, and everything else stays eligible.
- Verify it worked. Check that every source URL in the response sits on your allowlist (include case), or that none of them sit on an excluded domain (exclude case).
Expected: the result set obeys the filter exactly. A source outside the list means the policy was nested in the wrong place or the domain was written in the wrong form.
Variant phrasings
parallel ai search allowed domains
Same configuration: sourcepolicy with includedomains, apex domains only, nested under advanced_settings on the Search API.
parallel search restrict results to a domain
Use include_domains with that one domain to restrict results to it, or a domain/path prefix like example.com/blog to scope the filter to one section of a site (step 2, and the edge cases below).
exclude websites from parallel search results
That is the excludedomains path in step 3. Remember it is ignored when includedomains is also set.
parallel source policy docs
The rules that matter are the formatting rules in step 1 and the include-wins rule: one list per request, apex form, no wildcards.
Why the confusion happens
Include and exclude are not combined. Most failed attempts set both lists and expect both to apply; the include list wins and the exclude list is silently ignored. The second common failure is domain form: entries written with a scheme, a port, a www. prefix, or wildcard syntax do not match the way the writer expected.
Edge cases
- Domain/path prefixes like example.com/blog scope the filter to one section of a site, but path filters are unsupported in turbo mode; use fast, basic, or advanced mode instead.
- Long lists dilute the filter: up to 200 entries are allowed, but a short targeted list returns better results.
- Extensions as entries (.gov, .org, .co.uk) match broadly; use them only when the whole extension really is approved.
Tool compatibility
Parallel Search API (sourcepolicy under advancedsettings), and the same source policy feature on the Parallel Task, Monitor, and Responses APIs. Checked against the Parallel source policy docs below.
Related skills
- How to set approved domains in Parallel sourcepolicy with includedomains: https://vectle.com/skills/skl_TddjXt2VAwoiWIY9BoF3EA
- Parallel Search: unset mode silently defaults to advanced (slowest, priciest), set turbo or fast: https://vectle.com/skills/skl_KLG27Fqnjtl2NrDs8VhZfQ
Provenance
Built from the Parallel source policy docs: https://docs.parallel.ai/resources/source-policy