Audit candidate: Imbad0202/academic-research-skills

[xiaolai]

Manually queued from GitHub Trending (today) on 2026-05-21.

• Stars: 16210
• NL artifacts (path-pattern probe): 58
• Last push: 2026-05-20
• Vendor-default filter: OK

Source: trending-tab manual selection. The auditor-audit workflow fires when the audit-ready label is applied.

[github-actions]

Vocab drift advisory: 4 candidate(s). Report: auditor/audits/Imbad0202-academic-research-skills.vocab-drift.md on the auditor branch.

This is advisory only — vocabulary discipline is project-specific and per-repo. No PR will be filed for these findings.

[github-actions]

Audit Complete: Imbad0202/academic-research-skills

NL Score: 77/100 | Security: UNKNOWN | Artifacts: 56 | Strategy: batched

NL Audit: Imbad0202/academic-research-skills

NL Score Summary

Metric	Value
NL Score	77/100
Files scored	55
Agents	39
Commands	10
Skills / Config	6
Security	CLEAR
Threshold	70 (default)
Status	PASS

Score Breakdown by Suite

Suite	Files	Avg	Key Penalties
deep-research agents (14)	14	77	model missing (11/14), zero examples (14/14)
academic-paper-reviewer agents (7)	7	79	model missing (7/7), zero examples (5/7)
academic-pipeline agents (5)	5	75	model missing (5/5), orchestrator vague-cap −20
academic-paper agents (12)	12	88	model missing (12/12), but 10/12 have 2+ examples
shared agents (1)	1	86	model missing
commands (10)	10	60	no name:, no allowed-tools:, no empty-input handling
SKILL.md files (4)	4	88	version mismatches (bugs)
config / hooks (2)	2	85	—

Universal Issues

1. Missing model: frontmatter (36 / 39 agents, −5 each) — Only synthesis_agent, research_architect_agent, and report_compiler_agent declare model: inherit. Every other agent omits the field, leaving model routing to caller defaults.

2. Missing example blocks (23 / 39 agents, −15 each) — The deep-research and academic-paper-reviewer suites have no dedicated example sections. The academic-paper suite is notably better: 10 of 12 agents include 2+ inline examples (scored at 0-penalty).

3. Commands missing name: frontmatter (10 / 10, −25 each) — All 10 commands declare only description: and model:. The name: field is absent from every command file.

4. Commands missing allowed-tools: (10 / 10, −5 each) — No command specifies which tools it allows.

5. Commands missing empty-input handling (10 / 10, −10 each) — No command addresses invocation without arguments.

---

Security Scan

Risk Level: CLEAR

Pre-scan: 0 critical pattern matches, 0 high pattern matches (1 hook file, 130+ scripts, 0 MCP configs).

---

Full report: https://github.com/xiaolai/nlpm-for-claude/blob/main/auditor/audits/Imbad0202-academic-research-skills.md

No bugs found. Quality issues only, no PRs needed.