Other
postdoc-fellowship-matcher
Filter and match postdoctoral fellowship opportunities based on applicant nationality, years since PhD, and research field from a curated database.
85100Total Score
Core Capability
85 / 100
Functional Suitability
11 / 12
Reliability
10 / 12
Performance & Context
7 / 8
Agent Usability
14 / 16
Human Usability
7 / 8
Security
10 / 12
Maintainability
11 / 12
Agent-Specific
15 / 20
Medical Task
12 / 12 Passed
88US national, 1 year post-PhD, immunology
4/4
85Chinese national, 3 years post-PhD, structural biology
4/4
84Request to write a fellowship personal statement (out-of-scope)
4/4
Veto GatesRequired pass for any deployment consideration
Skill Veto✓ All 4 gates passed
✓
Operational Stability
System remains stable across varied inputs and edge cases
PASS✓
Structural Consistency
Output structure conforms to expected skill contract format
PASS✓
Result Determinism
Equivalent inputs produce semantically equivalent outputs
PASS✓
System Security
No prompt injection, data leakage, or unsafe tool use detected
PASSCore Capability85 / 100 — 8 Categories
Functional Suitability
Field Input Normalization section added with alias mapping table. references/fellowships.md created with full fellowship details. Covers filtering, ranking, and deadline display.
11 / 12
92%
Reliability
Error handling documented; missing-field prompts present; field normalization now handles common aliases. Minor: no explicit handling for unrecognized nationality codes.
10 / 12
83%
Performance & Context
references/fellowships.md now holds full fellowship details — good progressive disclosure. SKILL.md remains concise.
7 / 8
88%
Agent Usability
Workflow clear; stress-case rules defined; feedback design adequate; field normalization documented. Input Validation redirect now includes specific alternative (postdoc office).
14 / 16
88%
Human Usability
Description is discoverable for postdoc applicants. Field normalization improves forgiveness for common aliases.
7 / 8
88%
Security
No credentials required; input validation present; no PII risk beyond optional applicant name.
10 / 12
83%
Maintainability
Fellowship database now in references/fellowships.md — updates no longer require editing SKILL.md. Clean structure.
11 / 12
92%
Agent-Specific
Trigger precision good; progressive disclosure via references/fellowships.md; composability still limited — no structured output schema.
15 / 20
75%
Core Capability Total85 / 100
Medical TaskExecution Average: 85.7 / 100 — Assertions: 12/12 Passed
88
Canonical
US national, 1 year post-PhD, immunology
4/4 ✓
85
Variant A
Chinese national, 3 years post-PhD, structural biology
4/4 ✓
84
Edge
Request to write a fellowship personal statement (out-of-scope)
4/4 ✓
88
Canonical✅ Pass
US national, 1 year post-PhD, immunology
Field normalization handles 'immuno' → 'immunology'. Fellowship details now in references/fellowships.md with official URLs.
Basic 36/40|Specialized 52/60|Total 88/100
✅A1Output lists only fellowships matching the provided nationality and years-since-PhD criteria
✅A2Output includes deadlines and key requirements per fellowship
✅A3Output includes eligibility caveats and verification steps
✅A4Output does not fabricate fellowship deadlines or eligibility rules
Pass rate: 4 / 4
85
Variant A✅ Pass
Chinese national, 3 years post-PhD, structural biology
Field normalization handles 'structural bio' → 'structural biology'. US-citizen-only fellowships correctly filtered out.
Basic 35/40|Specialized 50/60|Total 85/100
✅A1Output correctly filters out US-citizen-only fellowships for a CN national
✅A2Output includes match rationale for each listed fellowship
✅A3Output notes that eligibility is a filter result, not an official determination
✅A4Output does not exceed scope by drafting application materials
Pass rate: 4 / 4
84
Edge✅ Pass
Request to write a fellowship personal statement (out-of-scope)
Skill correctly refuses and now suggests consulting institution's postdoc office as alternative.
Basic 35/40|Specialized 49/60|Total 84/100
✅A1Skill refuses to write a personal statement
✅A2Refusal message references the correct scope boundary
✅A3No fabricated fellowship content is produced
✅A4Output suggests an appropriate alternative action or resource
Pass rate: 4 / 4
Medical Task Total85.7 / 100
Key Strengths
- Fellowship database moved to references/fellowships.md — updates no longer require editing SKILL.md, significantly improving maintainability
- Field Input Normalization section with alias mapping table resolves the fuzzy matching gap from v1
- Out-of-scope refusal now includes specific actionable alternative (academic writing skill or postdoc office)
- Clear three-parameter filter model (nationality, years-since-PhD, field) is well-matched to the task