MedGaugeCapabilities / D · Medical practice
D.10 Research assistance Active
Given a research request, the agent produces a literature catalogue, a search strategy or data-access application material. Graded by checking the artifact fields and entries against the reference.
Evidence: none. No benchmark card lists this capability yet. No level is issued: methodology 0.1.0 sets no level thresholds.
- Domain
- D · Medical practice
- Evidence form
- program
- Veto constraints
- X.1 No fabricated evidenceX.2 Read-only input integrityX.4 Embedded instruction resistance
- Depends on
- A.6 Research evidence semanticsB.3 Tool invocationC.5 Pipeline orchestration
- Required by
- E.2E.5
- Internal task families
- traj_lit_ops
- External anchors
- —
- Version
- 0.1.0 · 0.1.0
Definition in detail
Research secretaries spend much of their time building literature catalogues, search strategies and data access applications. These artifacts must be complete and free of invented references.
Inputs a task gives
- A research request stating the question and scope
- Bibliographic records or search exports
- Database field descriptions
- A data access application template with its requirements
Artifact required
One of: a literature catalogue (record identifier, bibliographic fields, inclusion decision, reason); a search strategy (concept blocks, terms, operators, databases, limits); or completed application material (template fields filled from the supplied information).
How it is graded
A program checks artifact fields and entries against the reference: catalogue entries matched by identifier with their inclusion decisions, required strategy blocks and terms, and required application fields present and consistent with the supplied information. The one-shot vetoes X.1, X.2 and X.4 apply.
In scope
- Cataloguing and de-duplicating supplied records
- Structuring a search strategy from a research question
- Filling application material from supplied information
Out of scope
- Judging whether a citation supports a claim - graded under A.6 and D.6
- Running analysis pipelines - graded under C.5
Typical failures
Expected failure patterns; not measured rates.
- Adding references that are not in the supplied records
- Leaving duplicates that differ only in formatting
- Filling application fields with content the supplied information does not state
Benchmarks
No benchmark card lists this capability.
Grading references
No grading reference is mapped to this capability.