greenhouse-pipeline-search

Greenhouse pipeline search

Finds candidates across a pipeline (or across all pipelines) matching criteria the recruiter cares about, without them having to manually scroll through Greenhouse.

When this applies

Trigger when the user is looking for candidates that match some filter — a skill, years of experience, location, current stage, source, or a combination. If they name one specific person, that's greenhouse-candidate-screening instead. If they want you to review an entire unfiltered batch of resumes for a role (no specific filter, just "look at everyone"), that's greenhouse-resume-batch-review.

How to do it

1. Clarify scope if it's ambiguous: which job/pipeline (or "all"), and what the actual filter criteria are. If the user says something vague like "find good candidates," ask what "good" means for this search (specific skills? years of experience? something else?) rather than inventing criteria — a wrong guess wastes their time reviewing results that don't match what they meant.
2. Use search_pipeline_candidates for criteria-based search within a pipeline, or scan_pipeline_resumes when the match criteria live in resume content itself (e.g. specific technologies, certifications, or experience described in free text rather than structured fields).
3. If the user's criteria span multiple pipelines/jobs, run the search per job and combine — don't silently narrow to just one job because it's easier.
4. Rank or order the results in a way that's actually useful — e.g. best-match first, or by pipeline stage — rather than returning them in raw API order.

Output format

## Pipeline search — [criteria] in [job/pipeline scope]

Found N matching candidates:

| Candidate | Stage | Why they match |
|---|---|---|
| [name] | [stage] | [specific evidence from resume/profile, not a generic restatement of the criteria] |

If zero candidates match, say so plainly and suggest loosening the criteria — don't pad the response with near-misses unless you also label them clearly as near-misses.