Plugin packages of skills for journalists and researchers (verification, FOIA, AP style) plus dev, newsroom-ops and security toolkits for Claude Code and Codex.
A broad, actively maintained MIT collection with sensible safety text; pick the plugin package you need rather than everything.
This repo packages about 76 skills into installable plugins for Claude Code, with a legacy route that exposes the journalism skills to Codex. The journalism-core package has 16 skills: AP-style writing, AI-slop detoxing, source verification (SIFT, deepfake and C2PA checks), fact-check workflow, FOIA and NJ OPRA requests, Brazilian records requests, interview prep and transcription, data journalism, photo metadata with an exiftool-based Python script, editorial workflow and newsletter publishing. dev-toolkit adds 14 development skills (accessibility, ethical web scraping, minimal-code implementation, mobile debugging, Electron, Python pipelines), and newsroom-agent-ops has nine skills for unattended recurring work such as briefings, intake, delivery and deadline tracking. Other packages include pdf-playground, project-templates, an autocontext memory plugin, okf-wiki (scaffolds a markdown knowledge base with a validator and session-start hooks) and a security toolkit. Seventeen optional hooks are installed by hand, for example AP-style and source-attribution checks. SKILL.md files average about 13 KB, and many carry an untrusted-content boundary section telling the agent to treat retrieved text as data.
76 skills averaging 13 KB, with references in some, a handful of Python scripts and tests, 17 optional hooks, but no evals.
Counted from the repository file tree. How we score depth.
A large collection of markdown skills for journalism and development, plus a few local scripts: an exiftool-based photo metadata embedder, an OKF knowledge-base scaffolder with two optional session hooks, and PowerShell hooks that gate writes to risky files. The scripts we read make no network calls except an optional Playwright script that replays saved GitHub session cookies against github.com. Most of the repository's files were not available to us, so this is a partial review.
The repository (937 files) on Oct 29, 2026. Not a full review: 624 of 937 files were not downloaded (large repository; we reviewed the skill folders); it contains compiled binaries or archives we cannot read; we read 10 of 15 code files; it relies on code outside the repository: exiftool binary used by the photo-metadata script; Playwright Python package and the unseen gh-session-keepwarm.py used by gh-wiki-bootstrap.py; security-toolkit/test-fixtures/fake-mini-shai.tgz (archive).
NVIDIA SkillSpector static scan: 369 pattern hits (Tool Misuse 139, Agent Snooping 130, Dangerous Code Execution 34, Excessive Agency 26, Data Exfiltration 11, Rogue Agent 8). The 369 SkillSpector hits are mostly documentation: the --no-validate flag of the okf-wiki scaffolder in docs and tests, example code (requests.get, Sentry, LambdaTest, git reset --hard, credentials.json), and the two okf-wiki example hooks flagged as concealed executables, which only print the bundle index and block the first call once. Env-variable hits are in the okf-wiki tests that build subprocess environments. Large test files (test_okf_wiki.py, test_embed.py) and the two copies of validate.py were only searched, not read in full; most of the 937-file repository was not downloaded.
A review, not a certification. Skills change after we check them — read the files before installing.
MIT — commercial use allowed. A third-party notices file covers borrowed material.
Yes. The newsroom-agent-ops package includes an unattended-agent-runs skill built around explicit authority, durable state, bounded resources and recovery that avoids repeating external actions, though the other packages are interactive.
Five very detailed Claude Code skills for the academic pipeline: deep research, paper writing, peer review, revision and systematic-review screening.
A single research skill with an 8-phase pipeline, parallel search, source scoring, claim and citation checks, and report output in Markdown, HTML and PDF.
Research skill that searches Reddit, X, YouTube, Hacker News, Polymarket and more for the last 30 days, scores results by engagement, and writes a cited brief.
Academic research plugin with 15 skills; literature reviews check citations and quotes against OpenAlex, arXiv and Europe PMC.
A collection of about 76 skills and plugins for journalists, researchers and academics, plus developer and newsroom-operations toolkits, written for Claude Code and also usable with Codex.
Run /plugin marketplace add jamditis/claude-skills-journalism, then install a package such as journalism-core with /plugin install, and restart Claude Code. Choose packages individually; the README lists the commands for each.
No. The README says the 17 standalone hooks are optional and installed manually, and that installing a skill package does not enable them.
No. The source-verification and fact-check skills provide methods (SIFT, C2PA checks, verification trails) for the agent to follow, but the final editorial judgment stays with you.