agentleFS
Xuzhen-LiXuzhen-Li/literature-review-skills/AGENTS.md

literature-review-skills

AGENTS.md2 starsChanged 35 days agoMITView on GitHub

AGENTS.md is a guidance file for Codex and other agents. This repository provides 18 literature-review skills (stage instruction packs), shared/ rules, and install.sh. To install skills into ~/.cursor/skills, ~/.claude/skills, or ~/.agents/skills (which Codex reads), run install.sh with --target set to cursor, claude, agents, or all (default all). This file helps humans and agents locate skill paths, gates (checkpoints), and tests. Select the entry skill based on what the user already has (full mapping in docs/START_HERE.md): Below are the 18 skills…

The file

We split this AGENTS.md into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Keep SKILL.md frontmatter limited strictly to name and description.

    line 71

≈ “Common” is an estimate: at least 1 in 5 of the 41,581 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md

AGENTS.md is a guidance file for Codex and other agents. This repository provides 18 literature-review skills (stage instruction packs), shared/ rules, and install.sh. To install skills into ~/.cursor/skills, ~/.claude/skills, or ~/.agents/skills (which Codex reads), run install.sh with --target set to cursor, claude, agents, or all (default all). This file helps humans and agents locate skill paths, gates (checkpoints), and tests.

Start here

Select the entry skill based on what the user already has (full mapping in docs/START_HERE.md):

  • Topic only: review-project-setup.
  • Own papers or DOI list: review-project-setup. Export the collection as RIS (Zotero: Export Collection > RIS; for PDF folders, import into Zotero first to extract metadata) or build a CSV with a doi,title,year header, then run search-export-and-dedup with --source manual. Note that dedup.py only accepts .ris, .nbib, PubMed .txt, and .csv (plain-DOI .txt, raw PDF folders, and BibTeX are not supported).
  • Existing draft: review-writing (no project.yaml needed).
  • Near-final manuscript: journal-selection.
  • Decision letter: reviewer-response.
  • Unsure or mid-project: review-orchestrator.
  • No library account is needed. The default flow is AI-first: skills/search-export-and-dedup/scripts/api_search.py searches OpenAlex, Europe PMC, PubMed, Semantic Scholar, and arXiv (Crossref on request) and writes hit counts; skills/zotero-library-sync/scripts/find_oa_pdfs.py downloads legal OA PDFs; for every missed paper, the agent retries with its own browser or web-fetch tool (skills/zotero-library-sync/scripts/agent_retry.py: worklist lists the candidate OA URLs, and accept ingests the saved PDFs); skills/zotero-library-sync/scripts/please_fetch.py gives the user only the ranked list of essential papers still missing after that retry. Never use shadow libraries.
  • Optional extras: subscription databases (use_subscription_databases: true plus the optional institutional_access gate) and Zotero (use_zotero: true). Neither is required.

Repository layout

  • skills/<name>/SKILL.md
  • skills/<name>/references/
  • skills/<name>/scripts/
  • skills/<name>/tests/
  • shared/
  • install.sh
  • docs/DESIGN.md

Skills

Below are the 18 skills provided by this repository:

SkillPath
exemplar-review-analysisskills/exemplar-review-analysis/SKILL.md
figure-permissionsskills/figure-permissions/SKILL.md
journal-selectionskills/journal-selection/SKILL.md
literature-screeningskills/literature-screening/SKILL.md
manuscript-polishingskills/manuscript-polishing/SKILL.md
paper-deep-readingskills/paper-deep-reading/SKILL.md
review-figures-and-tablesskills/review-figures-and-tables/SKILL.md
review-orchestratorskills/review-orchestrator/SKILL.md
review-outline-builderskills/review-outline-builder/SKILL.md
review-project-setupskills/review-project-setup/SKILL.md
review-scoping-and-gapsskills/review-scoping-and-gaps/SKILL.md
review-section-drafterskills/review-section-drafter/SKILL.md
review-writingskills/review-writing/SKILL.md
reviewer-responseskills/reviewer-response/SKILL.md
search-export-and-dedupskills/search-export-and-dedup/SKILL.md
search-strategy-builderskills/search-strategy-builder/SKILL.md
submission-packageskills/submission-package/SKILL.md
zotero-library-syncskills/zotero-library-sync/SKILL.md

Workflow

Start each session from review-orchestrator (skills/review-orchestrator/SKILL.md) to decide what step comes next. Route to exactly one next skill per turn. Do not start a second skill in the same turn. Inspect project.yaml and LOG.md to check stage gates (topic_lock, search_lock, inclusion_lock, outline_lock, journal_lock, final_submit). An earlier unpassed gate always beats a later file that already exists.

Human gates

Gates are human checkpoints defined in shared/gates.md. A gate passes only when its key in project.yaml equals the YAML boolean true. Values like false, "true", yes, or missing keys do not pass. The agent may record progress but must never confirm a gate for the user. Every confirmation entry in LOG.md must end with | human.

Tests

python -m pytest -q

Open a terminal at the repository root and run pytest (the Python test runner) using the command above. CI runs the tests on several combinations of Python and pandas versions. Some skill tests require pandas.

Editing skills

Keep SKILL.md frontmatter limited strictly to name and description. The name must match the skill directory name (using lowercase, digits, and hyphens). Keep each SKILL.md under 500 lines; place long templates inside references/ (such as skills/<name>/references/). When a skill includes skills/<name>/scripts/, place matching tests under skills/<name>/tests/.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.