Academic Writing

target-journal-matcher

Matches a manuscript abstract to target journals using Tier 1/2/3 classification, NLP/clinical-trial/methodology-aware scoring, mandatory IF disclaimer, and open-access filter. Second polish: Python script rewritten — tier labels implemented, NLP field detection fixed, Cell penalized for clinical-trial papers, superconductor correctly routes to Nature/Science, IF disclaimer footer added to all output formats, --open-access CLI flag added.

86100Total Score

Core Capability

84 / 100

Functional Suitability

11 / 12

Reliability

9 / 12

Performance & Context

6 / 8

Agent Usability

13 / 16

Human Usability

7 / 8

Security

12 / 12

Maintainability

10 / 12

Agent-Specific

16 / 20

Medical Task

25 / 25 Passed

89CRISPR gene editing in stem cells (biology/medicine)

5/5

88Transformer NLP architecture for edge devices (AI/NLP)

5/5

85Minimal valid abstract length boundary (environmental science)

5/5

91Large multi-center RCT immunotherapy in NSCLC (clinical medicine)

5/5

88Room-temperature superconductor via quantum computing + materials screening (multidisciplinary)

5/5

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

PASS

Research Veto✓ PASS — Applicable

Dimension	Result	Detail
Scientific Integrity	PASS	Impact factor values sourced from bundled journals.json; no fabricated IFs, DOIs, PMIDs, or clinical outcome data.
Practice Boundaries	PASS	No diagnostic or prescriptive medical conclusions produced; skill is limited to journal recommendations.
Methodological Ground	PASS	No methodological fallacies; no ethical compliance requirements triggered by journal-matching task.
Code Usability	PASS	Script (main.py) runs successfully on Python 3.9. Classes AbstractAnalyzer, JournalDatabase, JournalMatchmaker are syntactically correct and produce output on all valid inputs.

Core Capability84 / 100 — 8 Categories

Functional Suitability

Tier 1/2/3 classification implemented in script output (all formats). NLP/CV field disambiguation fixed. Methodology penalty for clinical-trial vs basic-science journals. Multidisciplinary paradigm-shift detection added. --open-access filter added.

11 / 12

92%

Reliability

Fault Tolerance (3/4): short-abstract validation works; auto-creates default database if journals.json missing. Error Reporting (3/4): clear error messages for invalid inputs; no warning when result set is sparse (e.g., only 2 journals found for environmental topics). Recoverability (3/4): stateless CLI; each run is independent.

9 / 12

75%

Performance & Context

Token Cost (3/4): script output is concise. Execution Efficiency (3/4): fast runtime; no redundant computation; config in separate JSON files. Minor: no caching for repeated abstract analysis.

6 / 8

75%

Agent Usability

Mandatory IF disclaimer now embedded in all output formats. Tier-grouped output improves feedback design. Study design printed to console before recommendations.

13 / 16

81%

Human Usability

Discoverability (3/4): markdown output is clean and readable. Forgiveness (3/4): no destructive operations; errors are informative.

7 / 8

88%

Security

Full marks. No eval/exec on user input; CLI arguments properly handled; no credential exposure; no injection vectors.

12 / 12

100%

Maintainability

Score-to-tier mapping is a standalone function (easy to recalibrate). BASIC_SCIENCE_ONLY_JOURNALS set is externally updatable. Separate methodology penalty function.

10 / 12

83%

Agent-Specific

Tier labels enable progressive disclosure (Tier 1→2→3 decision tree). --open-access escape hatch added. Multidisciplinary always-include logic prevents missed high-value targets.

16 / 20

80%

Core Capability Total84 / 100

Medical TaskExecution Average: 88.2 / 100 — Assertions: 25/25 Passed

Canonical

CRISPR gene editing in stem cells (biology/medicine)

5/5 ✓

Variant A

Transformer NLP architecture for edge devices (AI/NLP)

5/5 ✓

Edge

Minimal valid abstract length boundary (environmental science)

5/5 ✓

Variant B

Large multi-center RCT immunotherapy in NSCLC (clinical medicine)

5/5 ✓

Stress

Room-temperature superconductor via quantum computing + materials screening (multidisciplinary)

5/5 ✓

Canonical

Pass

CRISPR gene editing in stem cells (biology/medicine)

Tier 1/2/3 labels now present in output (Tier 1: Cell, Nat Med, Nat Methods, Nat Biotechnol; Tier 1: Cell Res). IF disclaimer present. Basic science study design correctly detected — clinical journals appropriately deprioritized.

Basic 37/40|Specialized 52/60|Total 89/100

A1Output contains at least 3 journal recommendations

A2Returned journals are relevant to biology/methods fields (not environmental or physics)

A3Output includes Tier 1/2/3 classification as described in SKILL.md four-step workflow

A4Output does not guarantee acceptance at any recommended journal

A5Output includes caveat that IF values are approximate or potentially outdated per SKILL.md hard rule

Pass rate: 5 / 5

Variant A

Pass

Transformer NLP architecture for edge devices (AI/NLP)

NLP field now detected first (score highest). TACL (#1) and Computational Linguistics (#2) now lead recommendations. CV journals (TPAMI, IJCV) correctly deprioritized. Tier labels present. IF disclaimer present.

Basic 36/40|Specialized 52/60|Total 88/100

A1Output contains at least 3 journal recommendations

A2NLP-specific journals (TACL or Computational Linguistics) appear in recommendations for an NLP paper

A3Output includes Tier 1/2/3 classification as described in SKILL.md

A4Output does not guarantee acceptance at any recommended journal

A5Output includes caveat that IF values are approximate or potentially outdated

Pass rate: 5 / 5

Edge

Pass

Minimal valid abstract length boundary (environmental science)

Minimal environmental abstract returns Nature Climate Change (Tier 1) and Environmental S&T (Tier 2) — appropriate. Tier labels and IF disclaimer present.

Basic 35/40|Specialized 50/60|Total 85/100

A1Abstract shorter than 50 characters is correctly rejected with informative error message

A2Minimal valid abstract (60+ chars) returns at least 1 relevant journal

A3Output includes Tier 1/2/3 classification as described in SKILL.md

A4Output does not guarantee acceptance at any recommended journal

A5Output includes caveat that IF values are approximate or potentially outdated

Pass rate: 5 / 5

Variant B

Pass

Large multi-center RCT immunotherapy in NSCLC (clinical medicine)

Clinical trial design detected. Cell now receives 0.2× penalty (basic-science-only journal) and does not appear in top 5. Lancet (#1), JAMA (#2), NEJM (#3) correctly lead — all Tier 1. Tier labels and IF disclaimer present.

Basic 37/40|Specialized 54/60|Total 91/100

A1NEJM and The Lancet appear in recommendations for a large multi-center RCT

A2Cell is not ranked equal to or above NEJM/Lancet for a large RCT (methodology mismatch)

A3Output includes Tier 1/2/3 classification as described in SKILL.md

A4Output does not guarantee acceptance at any recommended journal

A5Output includes caveat that IF values are approximate or potentially outdated

Pass rate: 5 / 5

Stress

Pass

Room-temperature superconductor via quantum computing + materials screening (multidisciplinary)

Multidisciplinary field detected first (paradigm-shift + cross-disciplinary language). Nature (#1, Tier 1), Science (#2, Tier 1) correctly lead. Environmental S&T absent — chemistry field no longer triggered by superconductor abstract. npj Quantum Materials appears as physics/materials option. Tier labels and IF disclaimer present.

Basic 36/40|Specialized 52/60|Total 88/100

A1Environmental Science & Technology does not appear in recommendations for a quantum computing / superconductor paper

A2Nature or Science (multidisciplinary) appears in recommendations for a paradigm-shift discovery claim

A3Output includes Tier 1/2/3 classification as described in SKILL.md

A4Output does not guarantee acceptance at any recommended journal

A5Output includes caveat that IF values are approximate or potentially outdated

Pass rate: 5 / 5

Medical Task Total88.2 / 100

Key Strengths

Structured CLI interface with multiple output formats (table, json, markdown) and configurable --min-if / --max-if filtering enables realistic tier-scoped searches
Self-creating default journal database (auto-generates journals.json if missing) ensures operational stability without external setup steps
Short-abstract validation (50-char minimum) with informative error message correctly blocks malformed inputs
Well-organized Python codebase with separated configuration files (journals.json, fields.json, scoring_weights.json) enabling database updates without code changes
Configurable scoring weights file (scoring_weights.json) allows tunable matching behavior without rewriting logic