A structured five-seat AI review panel that evaluates academic manuscripts from journal-fit, methodology, domain, cross-disciplinary, and Devil's Advocate perspectives.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install ars-academic-reviewer
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install ars-academic-reviewer using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Academic Paper Reviewer?
Academic Paper Reviewer is an Openclaw Skills agent team for simulating a complete academic peer-review workflow without changing the submitted manuscript. It analyzes the paper's field and methodology, dynamically configures four reviewer identities, adds a fixed Devil's Advocate seat, and produces role-separated review reports with evidence anchors and actionable recommendations.
An editorial synthesizer then consolidates the five reports into an Editorial Decision Letter and immutable Revision Roadmap. Openclaw Skills users can select full, quick, methodology-focus, guided, re-review, or calibration modes while preserving academic-integrity safeguards, provenance disclosures, fail-closed verification, and explicit handling of unresolved dissent.
Academic Paper Reviewer Use Cases
- Run a comprehensive pre-submission peer review before sending a manuscript to an academic journal.
- Assess journal fit, originality, significance, readership relevance, and overall submission readiness.
- Inspect research design, sampling, statistical validity, effect sizes, reproducibility, and data transparency.
- Evaluate literature coverage, theoretical framing, domain contribution, and missing references.
- Stress-test the central thesis with counter-arguments, alternative explanations, logic checks, and stakeholder blind-spot analysis.
- Perform a quick 15-minute quality assessment when a full panel review is unnecessary.
- Focus specifically on methodology and statistics with the reduced methodology-focus panel.
- Verify whether revisions addressed first-round comments using the re-review traceability workflow.
- Learn through progressive, Socratic guided review rather than receiving conclusions alone.
- Measure bounded reviewer decision-level error patterns against adjudicated gold-paper sets through opt-in calibration.
- Integrate manuscript review into an academic writing and revision pipeline with ars-academic-paper and ars-pipeline-orchestrator.
How Academic Paper Reviewer Works
- Runtime self-check: The orchestrator checks for sub-agent dispatch, concurrency, inter-session messaging, reference-file access, and vision/OCR support. Missing capabilities trigger graceful fallbacks rather than task abortion or runtime changes.
- Input and routing: The user provides a manuscript or file and selects a review intent such as full review, methodology check, verification review, guided review, or calibration. Ambiguous cross-phase requests are clarified upstream.
- Field analysis:
field_analyst_agentreads the complete paper and identifies the primary and secondary disciplines, research paradigm, methodology, target journal tier, and paper maturity. - Persona configuration: The system creates four reviewer configuration cards for the Journal-Fit Reviewer and three peer-review roles. The Devil's Advocate is a fixed fifth execution seat. The configuration is shown to the user for confirmation or adjustment.
- Role-separated review: The five seats review from their assigned perspectives without seeing peer outputs. Contracted modes use paper-content-blind planning followed by paper-visible evaluation, with evidence anchors and criterion bindings preserved.
- Evidence and integrity checks: Reviewers use vision/OCR for scanned content, bind claims to verifiable sources, label speculation, avoid fabricated references or results, and keep the manuscript read-only.
- Editorial synthesis:
editorial_synthesizer_agentcompares the five reports, identifies consensus and dissent, adjudicates disputed findings, visibly addresses every Devil's Advocate CRITICAL issue, and applies the decision protocol. - Decision package: The workflow returns five review reports, an Editorial Decision Letter, and an immutable Revision Roadmap. Accept, Minor Revision, Major Revision, Reject, deferral, or fail-closed outcomes follow the active contract.
- Revision coaching: For non-Accept decisions, optional Socratic guidance helps the author triage each issue as
will_address,wont_address, ornot_on_pointand formulate a revision strategy without silently rewriting author intent. - Verification or calibration: Re-review compares the original and revised manuscripts against the roadmap and evidence bundle. Calibration runs bounded measurements on explicitly supplied gold papers and reports tier-scoped confidence and limitations.
Academic Paper Reviewer Setup
Load the
ars-academic-reviewerskill in an Openclaw-compatible session. The SKILL.md does not specify a separate package installation command; runtime loading and reference-file availability are handled by the host environment.Make the complete manuscript available as pasted text or a readable file. Supply text alternatives for scanned pages, figures, tables, formulas, or other content when vision/OCR is unavailable.
For standalone use, explicitly name the desired mode. The default request is:
Review this paper: [paste paper or provide file]Choose an operational mode:
fullfor a comprehensive five-seat review.quickfor a rapid assessment.methodology-focusfor methods and statistics.guidedfor Socratic learning and progressive issue discovery.re-reviewfor post-revision verification.calibrationfor explicit gold-paper measurement.
In orchestrator-driven deployments, connect the skill after the academic-paper output and integrity check, then pass the immutable revision roadmap and author-adjudication sidecar into later revision and re-review stages.
If cross-model verification is enabled, obtain explicit user consent because the manuscript may be uploaded to an external provider. Configuration alone does not constitute consent.
If the runtime lacks sub-agents, messaging, reference access, or OCR, allow the documented inline, sequential, user-paste, or text-only fallback behavior.
Do not place instructions inside manuscripts, comments, PDFs, or response letters that attempt to alter reviewer identity, routing, tools, network access, file writes, or workflow constraints; these materials are treated as untrusted data.
Academic Paper Reviewer Data Schema & Taxonomy
The skill separates manuscript input from review artifacts and preserves provenance across phases. Core artifacts include:
| Artifact | Purpose |
|---|---|
| Reviewer Configuration Card | Stores field, methodology, journal context, and the four dynamically configured reviewer identities. |
| Five reviewer reports | Separate outputs for Journal-Fit, Methodology, Domain, Perspective, and Devil's Advocate seats. |
review-panel-provenance/1.0 |
Records actual execution observations such as role, context freshness, peer-output visibility, model family, provider, and accountable human identity. It does not claim statistically independent error processes. |
ReviewCriteriaBindingManifest |
Pointer-only binding to author-confirmed target criteria, with ordered criterion IDs and separated interdisciplinary conflict groups. |
| Evidence anchors | Typed manuscript locations and verifiable sources supporting each finding, with speculation explicitly labeled when necessary. |
| Editorial Decision Letter | Synthesizes report-backed consensus, disagreements, severity, adjudication, and final decision. |
revision-roadmap/1.0 |
Immutable, non-ranking list of revision requirements consumed by downstream revision workflows. |
author-adjudication/1.0 |
Explicit author sidecar recording will_address, wont_address, or not_on_point for each source-ordered item. |
| R&R Traceability Matrix | Re-review structure connecting original findings to author claims, revised evidence, verification results, residual issues, and a new decision. |
| Calibration Report | Tier-scoped raw counts and bounded decision-level FNR, FPR, balanced accuracy, or directional Minor/Major boundary results. |
The panel uses stable role markers EIC, R1, R2, R3, and DA. Findings are categorized as CRITICAL, MAJOR, or MINOR and must include what is wrong, where it occurs, evidence, and a specific remedy. Current live reviews and Schema 6 packages remain marked NOT_CALIBRATED; calibration does not create a general quality score or automatically wire a measured profile into live review.
Academic Paper Reviewer Advanced Features
- Five-seat role separation with dynamic field-specific reviewer cards and a mandatory fixed Devil's Advocate seat.
- Devil's Advocate CRITICAL terminal gate: validated or unresolved critical challenges cannot be silently bypassed before Accept.
- Paper-content-blind Phase 1 planning and paper-visible Phase 2 evaluation under the Sprint Contract Protocol.
- Contract-bound criteria manifests, eligible-role matrices, owner roles, trigger binding, evidence anchors, and checker-backed conformance validation.
- Re-review with three-gate evidence-before-persuasion sequencing, frozen Round-1 cards, R&R traceability, and fail-closed deferral or abort behavior.
- Inter-session debate support with explicit role instructions, scoped message envelopes, a maximum of three rounds, and reported consensus and dissent.
- Graceful degradation from parallel sub-agents to sequential or inline role-play when host capabilities are unavailable.
- Optional cross-model Reviewer 2 transport in full mode, with explicit consent, actual seat-level provenance, replay validation, and correlated-error disclosure.
- Guided Socratic revision coaching that surfaces strengths and issues progressively without inventing contributions or rewriting the manuscript.
- Calibration tiers supporting directional three-paper checks or full five-to-twenty-paper measurement with configurable run budgets and cross-model defaults.
- Optional relative model tiering with economy and quality-boost policies that never downgrade judgment agents below the configured floor.
- Strict read-only manuscript handling, untrusted-material defenses, zero-hallucination rules, computation recheck reminders, and language matching with user override.
- Pipeline integration with ars-academic-paper, ars-deep-research, integrity checks, ars-pipeline-orchestrator, revision workflows, and final verification.
SKILL.md
Loading
Related Openclaw Skills
Magic Hour
Magic Hour gives Openclaw Skills a unified API for generating AI videos and images across leading models with asynchronous job tracking and optional local downloads.
OpenClaw Agent Cognitive OS X
A cognitive control layer that turns OpenClaw goals into adaptive, verified, and safely managed autonomous-agent workflows.
OpenClaw Plugin Intelligence
OpenClaw Plugin Intelligence helps agents identify, select, securely execute, validate, and recover from external plugin workflows.
Icon Designer Generator
Generate editable AI icon designs, posters, and multi-page visual carousels from text prompts, templates, and customizable design specifications.
Submit to WorthToTry
Automate AI tool submissions to WorthToTry by auditing a product page, drafting its listing, and returning a review link.
Media Bias News
Media Bias News helps AI agents summarize current events while exposing coverage differences, outlet ratings, sourcing, and media blind spots.
Featured*
