alterlab-deep-research — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited alterlab-deep-research (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Universal deep research tool — a domain-agnostic 13-agent team for rigorous academic research on any topic. v2.3 adds systematic review mode (PRISMA-compliant with optional meta-analysis), Socratic convergence criteria, and post-research literature monitoring.
Minimal command:
Research the impact of AI on higher education quality assuranceSocratic mode:
Guide my research on the impact of declining birth rates on private universities
引導我的研究：少子化對私立大學的影響
幫我釐清我的研究方向，我對高教品保有興趣但還不太確定Execution:
English: research, deep research, literature review, systematic review, meta-analysis, PRISMA, evidence synthesis, fact-check, methodology, APA report, academic analysis, policy analysis, guide my research, help me think through, monitor this topic, set up alerts
Türkçe: araştırma, derinlemesine araştırma, literatür taraması, literatür incelemesi, sistematik derleme, meta-analiz, kanıt sentezi, doğruluk kontrolü, araştırmamı yönlendir, araştırma sorumu netleştir, araştırma konusu
繁體中文: 研究, 深度研究, 文獻回顧, 文獻探討, 系統性回顧, 後設分析, 證據綜整, 事實查核, 研究方法, 學術分析, 政策分析, 引導我的研究, 幫我釐清, 監測這個主題, 設定追蹤
Activate socratic mode when the user's intent matches any of the following patterns, regardless of language. Detect meaning, not exact keywords.
Intent signals (any one is sufficient):
Default rule: When intent is ambiguous between socratic and full, prefer `socratic` — it is safer to guide first than to produce an unwanted report. The user can always switch to full later.
Example triggers (illustrative, not exhaustive): "guide my research", "help me think through", 「引導我的研究」「幫我釐清」, or equivalent in any language
| Scenario | Use Instead |
|---|---|
| Writing a paper (not researching) | alterlab-paper-writer |
| Reviewing a paper (structured review) | alterlab-paper-reviewer |
| Full research-to-paper pipeline | alterlab-research-pipeline |
| Auditing whether the references in an existing bibliography exist or are retracted | alterlab-citation-verifier |
| Pooling already-extracted effect sizes (random-effects model, forest/funnel plots, Egger's test) in Python or R | alterlab-meta-analysis |
In Claude Code, dual-reviewer PRISMA screening of hundreds of records, a literature map, or a claim stress-test with independent agents is packaged as a workflow (/alterlab-workflows:systematic-review-screening, literature-map, claim-stress-test; see alterlab-research-workflows) — offer it before hand-building a fan-out.
| Your Situation 你的狀況 | Recommended Mode |
|---|---|
| Vague idea, need guidance / 有模糊想法，需要引導 | socratic |
| Clear RQ, need comprehensive research / 有明確 RQ，需要完整研究 | full |
| Need a quick brief (30 min) / 需要快速摘要 | quick |
| Have a paper to evaluate before citing / 有論文需要評估 | review |
| Need literature review for a topic / 需要文獻回顧 | lit-review |
| Need to verify specific claims / 需要查核特定事實 | fact-check |
| Need systematic review / meta-analysis / 系統性回顧或後設分析 | systematic-review |
Not sure? Start with socratic — it will help you figure out what you need. 不確定？先用 socratic 模式——它會幫你釐清你需要什麼。
| # | Agent | Role | Phase |
|---|---|---|---|
| 1 | research_question_agent | Transforms vague topics into precise, FINER-scored research questions with scope boundaries | Phase 1, Socratic Layer 1 |
| 2 | research_architect_agent | Designs methodology blueprint: paradigm, method, data strategy, analytical framework, validity criteria | Phase 1 |
| 3 | bibliography_agent | Systematic literature search, source screening, annotated bibliography in APA 7.0 | Phase 2 |
| 4 | source_verification_agent | Fact-checking, source grading (evidence hierarchy), predatory journal detection, conflict-of-interest flagging | Phase 2 |
| 5 | synthesis_agent | Cross-source integration, contradiction resolution, thematic synthesis, gap analysis | Phase 3 |
| 6 | report_compiler_agent | Drafts complete APA 7.0 report (Title -> Abstract -> Intro -> Method -> Findings -> Discussion -> References) | Phase 4, 6 |
| 7 | editor_in_chief_agent | Q1 journal editorial review: originality, rigor, evidence sufficiency, verdict (Accept/Revise/Reject) | Phase 5 |
| 8 | devils_advocate_agent | Challenges assumptions, tests for logical fallacies, finds alternative explanations, confirmation bias checks | Phase 1, 3, 5, Socratic Layer 2, 4 |
| 9 | ethics_review_agent | AI-assisted research ethics, attribution integrity, dual-use screening, fair representation | Phase 5 |
| 10 | socratic_mentor_agent | Q1 journal editor persona; guides research thinking through Socratic questioning across 5 layers | Socratic Mode (Layer 1-5) |
| 11 | risk_of_bias_agent | Assesses risk of bias using RoB 2 (RCTs) and ROBINS-I (non-randomized); traffic-light visualization | Systematic Review (Phase 2) |
| 12 | meta_analysis_agent | Designs and executes meta-analysis or narrative synthesis; effect sizes, heterogeneity, GRADE | Systematic Review (Phase 3) |
| 13 | monitoring_agent | Post-research literature monitoring: digests, retraction alerts, contradictory findings detection | Optional (post-pipeline) |
See references/mode_selection_guide.md for the detailed guide.
User Input
|
+-- Already have a clear research question?
| +-- Yes --> Need PRISMA-compliant systematic review / meta-analysis?
| | +-- Yes --> systematic-review mode
| | +-- No --> Need a full report?
| | +-- Yes --> full mode
| | +-- No --> Only need literature?
| | +-- Yes --> lit-review mode
| | +-- No --> quick mode
| +-- No --> Want to be guided through thinking?
| +-- Yes --> socratic mode
| +-- No --> full mode (Phase 1 will be interactive)
|
+-- Already have text to review? --> review mode
+-- Only need fact-checking? --> fact-check modeThe standard pipeline runs six phases: 1. Scoping (RQ Brief + Methodology Blueprint, Checkpoint 1) → 2. Investigation (Source Corpus + verified/graded sources) → 3. Analysis (Synthesis + gap analysis, Checkpoint 2) → 4. Composition (full APA 7.0 draft) → 5. Review (editorial + ethics + Checkpoint 3, parallel) → 6. Revision (final report, max 2 loops).
Full agent-by-agent flow diagram with every per-phase output and checkpoint criterion: see references/orchestration_workflows.md (Standard Orchestration Workflow section).
Core principle: From the perspective of a Q1 international journal editor-in-chief, guide users to clarify their research questions through Socratic questioning. Never give direct answers; instead, use follow-up questions to help users think through the issues themselves.
See agents/socratic_mentor_agent.md for the detailed agent definition. See references/socratic_questioning_framework.md for the questioning framework.
Socratic mode advances through five layers: 1. Problem Framing → 2. Methodology Reflection → 3. Evidence Design → 4. Critical Self-Examination → 5. Significance & Contribution. Each layer pairs the socratic_mentor_agent with a supporting agent (research_question_agent at Layer 1; devils_advocate_agent at Layers 2 and 4), extracts [INSIGHT: ...] markers each round, and requires ≥2 dialogue rounds (Layer 5 ≥1) before advancing. Layer 5 compiles all INSIGHTs into a Research Plan Summary that hands off to alterlab-paper-writer (plan mode).
Full layer-by-layer flow with the exact guiding questions per layer: see references/orchestration_workflows.md (Socratic Mode Layer Flow section).
full mode (see Failure Paths F6)Full PRISMA-compliant systematic literature review with optional meta-analysis. This mode extends the standard 6-phase pipeline with specialized agents for risk of bias assessment (RoB 2, ROBINS-I) and quantitative synthesis.
See agents/risk_of_bias_agent.md and agents/meta_analysis_agent.md for detailed agent definitions. See references/systematic_review_toolkit.md for the Cochrane/PRISMA/GRADE reference guide.
This mode reuses the six phases but with PRISMA-specific outputs: 1. Scoping produces a PICOS RQ and a full PRISMA-P protocol (not just an RQ); 2. Investigation adds a PRISMA 2020 flow diagram and a risk_of_bias_agent RoB assessment; 3. Analysis runs meta_analysis_agent (effect sizes, heterogeneity, GRADE) or SWiM narrative synthesis alongside thematic synthesis; 4. Composition emits a 27-item PRISMA 2020 report with study-characteristics, RoB, forest-plot, and GRADE tables; 5. Review and 6. Revision mirror the standard pipeline.
Full agent-by-agent PRISMA flow with every per-phase output: see references/orchestration_workflows.md (Systematic Review Mode Pipeline section).
| Mode | Agents Active | Output | Word Count |
|---|---|---|---|
full (default) | All 9 core (excluding socratic_mentor, risk_of_bias, meta_analysis, monitoring) | Full APA 7.0 report | 3,000-8,000 |
quick | RQ + Biblio + Verification + Report | Research brief | 500-1,500 |
review | Editor + Devil's Advocate + Ethics | Reviewer report on provided text | N/A |
lit-review | Biblio + Verification + Synthesis | Annotated bibliography + synthesis | 1,500-4,000 |
fact-check | Source Verification only | Verification report | 300-800 |
socratic | Socratic Mentor + RQ + Devil's Advocate | Research Plan Summary (INSIGHT collection) | N/A (iterative) |
systematic-review | RQ + Architect + Biblio + Verification + RoB + Meta-Analysis + Synthesis + Report + Editor + Ethics + DA | Full PRISMA 2020 report + forest plot data + GRADE table | 5,000-15,000 |
See references/failure_paths.md for all failure scenarios, trigger conditions, and recovery strategies across all modes.
Key failure path summary:
| Failure Scenario | Trigger Condition | Recovery Strategy |
|---|---|---|
| RQ cannot converge | Phase 1 / Layer 1 exceeds multiple rounds while still vague | Provide 3 candidate RQs or suggest lit-review |
| Insufficient literature | bibliography_agent finds < 5 sources | Expand search strategy, alternative keywords |
| Methodology mismatch | RQ type misaligned with method capability | Return to Phase 1, suggest 3 alternative methods |
| Devil's Advocate CRITICAL | Fatal logical flaw discovered | STOP, explain the issue, require correction |
| Ethics BLOCKED | Serious ethical issue | STOP, list issues and remediation path |
| Socratic non-convergence | > 10 rounds without convergence | Suggest switching to full mode |
| User abandons mid-process | Explicitly states they don't want to continue | Save progress, provide re-entry path |
| Only Chinese-language literature | English search returns empty | Switch to Chinese academic databases |
After any research mode is complete, users can optionally activate the monitoring_agent to set up post-research literature monitoring. This is not part of the main pipeline — it is an auxiliary capability triggered on demand.
See agents/monitoring_agent.md for the detailed agent definition. See references/literature_monitoring_strategies.md for platform-specific setup guides.
Trigger: "monitor this topic", "set up alerts", "track new publications on this"
Capabilities:
Input: Completed bibliography + search strategy from any research mode Output: Monitoring configuration + digest template (markdown)
Limitation: The monitoring agent produces configurations and templates for the user to act on. It cannot run autonomous background monitoring.
After research is complete, the following materials can be handed off to alterlab-paper-writer:
Trigger: User says "now help me write a paper" or "write a paper based on this"
alterlab-paper-writer's intake_agent will automatically detect available materials and skip redundant steps:
See examples/handoff_to_paper.md for a detailed handoff example.
See alterlab-research-pipeline/SKILL.md for the complete workflow.
| Agent | Definition File |
|---|---|
| research_question_agent | agents/research_question_agent.md |
| research_architect_agent | agents/research_architect_agent.md |
| bibliography_agent | agents/bibliography_agent.md |
| source_verification_agent | agents/source_verification_agent.md |
| synthesis_agent | agents/synthesis_agent.md |
| report_compiler_agent | agents/report_compiler_agent.md |
| editor_in_chief_agent | agents/editor_in_chief_agent.md |
| devils_advocate_agent | agents/devils_advocate_agent.md |
| ethics_review_agent | agents/ethics_review_agent.md |
| socratic_mentor_agent | agents/socratic_mentor_agent.md |
| risk_of_bias_agent | agents/risk_of_bias_agent.md |
| meta_analysis_agent | agents/meta_analysis_agent.md |
| monitoring_agent | agents/monitoring_agent.md |
| Reference | Purpose | Used By |
|---|---|---|
references/apa7_style_guide.md | APA 7th edition quick reference | report_compiler, editor_in_chief |
references/source_quality_hierarchy.md | Evidence pyramid + grading rubric | source_verification, bibliography |
references/methodology_patterns.md | Research design templates | research_architect |
references/logical_fallacies.md | 30+ fallacies catalog | devils_advocate |
references/ethics_checklist.md | AI disclosure, attribution, dual-use | ethics_review |
references/interdisciplinary_bridges.md | Cross-discipline connection patterns | synthesis, research_architect |
references/socratic_questioning_framework.md | 6 types of Socratic questions + 30+ prompt patterns | socratic_mentor |
references/failure_paths.md | 12 failure scenarios with triggers and recovery paths | all agents |
references/mode_selection_guide.md | Mode selection flowchart and comparison table | orchestrator |
references/orchestration_workflows.md | Detailed 6-phase, Socratic 5-layer, and systematic-review flow diagrams | orchestrator |
references/irb_decision_tree.md | IRB decision tree (US tiers) + Taiwan process + HE quick reference; Turkey via alterlab-tr-research-ethics | ethics_review, research_architect |
references/equator_reporting_guidelines.md | EQUATOR reporting guideline mapping | research_architect, report_compiler |
references/preregistration_guide.md | Preregistration decision tree + platforms + checklist | research_architect |
references/systematic_review_toolkit.md | Cochrane Handbook v6.5, PRISMA 2020, RoB 2, ROBINS-I (2016 and V2), I² guide, GRADE, protocol registration | risk_of_bias, meta_analysis, bibliography, report_compiler |
references/literature_monitoring_strategies.md | Google Scholar alerts, PubMed alerts, RSS feeds, Retraction Watch, citation tracking, monitoring cadence | monitoring_agent |
| Template | Purpose |
|---|---|
templates/research_brief_template.md | Quick mode output format |
templates/literature_matrix_template.md | Source x Theme analysis matrix |
templates/evidence_assessment_template.md | Per-source quality assessment card |
templates/preregistration_template.md | OSF standard 21-item preregistration template |
templates/prisma_protocol_template.md | PRISMA-P 2015 systematic review protocol template |
templates/prisma_report_template.md | PRISMA 2020 systematic review report template (27 items) |
| Example | Demonstrates |
|---|---|
examples/exploratory_research.md | Full 6-phase pipeline walkthrough |
examples/systematic_review.md | PRISMA-style literature review |
examples/policy_analysis.md | Applied comparative policy research |
examples/socratic_guided_research.md | Complete Socratic mode multi-turn dialogue (12 rounds) |
examples/handoff_to_paper.md | alterlab-deep-research full mode handoff to alterlab-paper-writer |
examples/review_mode.md | Review mode: 3-agent review pipeline for policy recommendation text |
examples/fact_check_mode.md | Fact-check mode: source verification of HEI claims with per-claim verdicts |
Follows the user's language. Academic terminology kept in English. Socratic mode uses natural conversational style.
Unified definitions to prevent inconsistency across agents:
| Concept | Definition | Applies To |
|---|---|---|
| Peer-reviewed | Published in a journal with formal peer review process (editorial review alone does not qualify). Conference proceedings count only if explicitly peer-reviewed | bibliography_agent, source_verification_agent |
| Currency Rule | Default: published within 5 years. Override by domain: CS/AI = 3 years, History/Philosophy = 20 years, Law = depends on jurisdiction changes. Seminal works exempt regardless of age | bibliography_agent, ethics_review_agent |
| CRITICAL severity | Issue that, if unresolved, would invalidate a core conclusion or constitute academic misconduct. Requires immediate resolution before pipeline can proceed | All agents |
| Source Tier | tier_1 = top-quartile peer-reviewed journal; tier_2 = other peer-reviewed; tier_3 = academic but not peer-reviewed; tier_4 = grey literature | bibliography_agent, source_verification_agent |
| Minimum Source Count | full = 15+, quick = 5-8, lit-review = 25+, systematic-review = all eligible (no limit), fact-check = 3+ per claim | bibliography_agent |
| Verification Threshold | 100% of sources existence-checked before use (verify_citations.py verdicts, WebSearch fallback) + independent WebSearch spot-check of ≥50% of sources (all tier_3/tier_4 first) | bibliography_agent, source_verification_agent, ethics_review_agent |
Cross-Skill Reference: See shared/handoff_schemas.md for inter-stage data exchange formats.This skill is domain-agnostic but composes with other AlterLab skills:
alterlab-deep-research + alterlab-paper-2-web -> Disseminate research as a website / video / poster
alterlab-deep-research + alterlab-paper-writer -> Full research-to-publication pipeline
alterlab-deep-research (socratic) + alterlab-paper-writer (plan) -> Guided research + paper planning
alterlab-deep-research (systematic-review) + alterlab-paper-writer -> PRISMA systematic review paper
alterlab-deep-research -> alterlab-research-pipeline -> End-to-end research → review → finalize workflowCurrent: v2.4 (see frontmatter metadata.version). Latest change (2026-09): existence verification aligned across agents (every source is checked with alterlab-citation-verifier before use; no "unverifiable but kept" bucket), the Socratic stopping rules consolidated to one consistent set, and reporting standards brought current (Cochrane Handbook v6.5, ROBINS-I V2, CONSORT 2025, SPIRIT 2025, TRIPOD+AI). v2.3 added the systematic-review mode (PRISMA 2020 pipeline with risk_of_bias_agent and meta_analysis_agent) and the optional post-pipeline monitoring_agent, bringing the team to 13 agents and 7 modes.
Part of the AlterLab Academic Skills suite.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.