screening agent passed a candidate whose resume said seeking part-time for a full-time role - the constraint was...

screening agent passed a candidate whose resume said seeking part-time for a full-time role - the constraint was prose, not a field

TL;DR

The part-time requirement was written in the resume's summary paragraph, not in any field, so the field-matching agent never saw it. Extract constraints from prose explicitly: ask the agent to list work-arrangement requirements from the full text before scoring. Fields catch the structured; prose needs reading.

screening agent passed part-time-seeking candidate for full-time role - constraint was prose

Steps

1. Confirm the miss: find the prose constraint in the resume the agent scored. Expected: The sentence is there, plain as day.

2. Add a constraint-extraction step before scoring: list work arrangement, location, and availability requirements from the full text. Expected: Constraints become explicit.

3. Gate on hard constraints: part-time vs full-time mismatches fail before scoring. Expected: Mismatches never pass.

4. Re-screen recent passes for prose constraints the agent missed. Expected: Other misses surface.

5. Keep a constraint checklist per role (full-time, on-site, clearance) that the extraction must fill. Expected: Nothing structural is left to prose luck.

When to use

• resumes state constraints in prose
• passed candidates cannot take the role
• screening matches fields only

When not to use

• the constraint is genuinely ambiguous (escalate to a human)
• the role is flexible on arrangement (then it is not a hard gate)
• the resume has no such prose (nothing to extract)

Tool compatibility

• any screening agent
• full-text resume parsing

Variant phrasings

• "prose constraint missed screening"
• "part-time passed for full-time role"
• "resume prose requirement ignored"

Why it happens

Resumes are documents, not forms: the important facts often live in sentences, not fields. An agent that matches structured fields against structured requirements is blind to everything the candidate wrote in prose. The constraint was visible to any human reader; the agent never read that way.

Edge cases

• prose constraints can be hedged ('open to part-time'); teach the agent to distinguish firm from flexible
• the extraction step adds latency; run it once and cache the constraints
• candidates sometimes bury deal-breakers; the checklist should prompt for the role's known ones

Provenance

Resolved from the public thread: https://vectle.com/posts/pstY6bl3YJLtRsKsRMX4R7bg