llms.txt@docs git:20261028.516da0dA

Immutable. This exact content never changes and is served at /api/v1/blob/55880f83233a418f.

# Agent skills for journalism
# https://skills.amditis.tech

## What this is

A collection of standards-based agent skills and Claude Code workflow hooks for journalists, researchers, academics, and media professionals. The skills provide specialized tools and workflows for compatible agents, including Claude Code and Codex. Claude-specific commands, agents, and hooks remain scoped to Claude Code.

Source: https://github.com/jamditis/claude-skills-journalism

## Who made it

Joe Amditis, associate director of operations, Center for Cooperative Media, Montclair State University.
Personal site: https://jamditis.com

## Skills (76 total)

### Journalism and verification
- source-verification, SIFT method, verification trails, source credibility, deepfakes/C2PA
- foia-requests, drafting and tracking public records requests
- brazil-records-requests, Brazilian public records requests under the Access to Information Law (LAI)
- fact-check-workflow, claim verification, evidence chains
- crisis-communications, breaking news protocols, rapid verification
- social-media-intelligence, OSINT, account analysis, social media verification

### Reporting and editorial
- interview-prep, research, question development, background checks
- interview-transcription, recording notes, quote handling, attribution
- meeting-transcript-actions, turn verified meeting transcripts into attributed notes, decisions, actions, and unresolved questions
- newsroom-style, AP Style enforcement and editorial standards
- data-journalism, data analysis and data-driven storytelling
- editorial-workflow, assignment tracking, story calendars
- story-pitch, pitch templates and pitch-writing workflows
- photo-metadata, embed caption, byline, credit, alt text, keywords, and license into a photo's IPTC/EXIF/XMP metadata for wire and archive use

### Academic and research
- academic-writing, research writing, citation, lit review
- digital-archive, archive building and management
- web-archiving, Wayback, Archive.today, URL preservation
- content-access, Unpaywall, CORE, open access tools
- page-monitoring, change detection and alerts
- free-apis-catalog, curated free-API catalog with sunset/currency notes

### Writing quality
- ai-writing-detox, eliminate AI writing patterns from your work
- newsletter-publishing, email newsletters and subscriber workflows

### Development
- surface-completeness, shared UI coverage, design compliance, and current documentation
- vibe-coding, AI-assisted development for non-engineers
- one-way-door, flag irreversible architectural decisions
- electron-dev, Electron app patterns
- ponytail-dev, minimal-code implementation with proportional checks
- python-pipeline, data pipeline workflows
- web-scraping, content extraction
- zero-build-frontend, no-build web apps
- mobile-debugging, Eruda, vConsole, remote debug
- accessibility-compliance, WCAG, alt text, a11y
- web-ui-best-practices, signs of craft in web UI
- claude-md-updater, propose scoped project instruction updates from a session
- context-engineering-fundamentals, preserve instructions, evidence, and state in long agent sessions
- director, explicit top-tier direction through workers configured in project instructions

### Security
- security-checklist, pre-deployment security audit
- secure-auth, authentication patterns
- api-hardening, API security
- private-secret-scanning, pinned local Gitleaks scans for staged changes, push ranges, and full history in private repositories with redacted reports
- supply-chain-hardening, npm/bun install-time cooldown plus sandboxed pre-install scan; defends against Mini Shai-Hulud-class supply-chain worms (paired with /security-toolkit:hotpatch slash command)

### Project templates
- goal-progress-bar, evidence-based weighted progress from a JSON tracker
- project-memory, generate project instruction files for new projects
- project-retrospective, generate LESSONS.md files
- template-selector, pick the right project instruction or LESSONS.md template

### Newsroom agent operations
- unattended-agent-runs, recurring work with leases, checkpoints, recovery, delivery checks, and resource limits
- scheduled-briefing, source-backed phone-readable newsroom briefings
- deadline-and-reply-tracking, due dates, reply windows, waiting states, and next actions
- email-request-intake, allowlisted email intake with untrusted-content controls
- document-delivery, editable document and PDF delivery with inspection and readback
- chat-mention-triage, team-chat classification without duplicate replies
- record-reconciliation, canonical task, contact, and interaction records with privacy controls
- multi-agent-coordinator, durable worker coordination with watches, gates, and cleanup
- issue-pool-progression, claimed issue progression through merge, deployment, and live verification

### Documents and explainers
- pdf-design, design and edit professional PDF reports and proposals
- pdf-playground, document design (proposals, reports)
- web-design-picker, create distinct website directions, a client review picker, asset catalog, previews, QA reports, and delivery packages
- visual-explainer, generate self-contained HTML pages that visually explain systems
- okf-wiki, scaffold an Open Knowledge Format knowledge base (one-concept-per-file markdown, validator, session-start orientation hooks)

### Video
- video-download, download authorized video from supported social platforms
- video-transcribe, transcribe audio and video with local Whisper workflows
- video-frames, extract and inspect representative video frames
- video-dashboard, build a local dashboard from transcript and frame analysis

### Workflow patterns (superjawn)
- using-superjawn, establishes how to find and use skills
- brainstorming, required before creative work
- writing-plans, multi-step task planning before touching code
- writing-skills, creating or editing skills
- systematic-debugging, bug triage before proposing fixes
- subagent-driven-development, parallel task execution in one session
- dispatching-parallel-agents, independent tasks across 2+ agents
- executing-plans, run a written implementation plan
- requesting-code-review, verify completed work before merging
- receiving-code-review, handle review feedback effectively
- verification-before-completion, confirm work is complete before committing
- finishing-a-development-branch, decide how to land completed work
- using-git-worktrees, isolate feature work from current workspace

## Hooks (17 workflow hooks)

Hooks run automatically at specific workflow events. Most are non-blocking warnings. `one-way-door-check` exits 2 to prevent irreversible architectural decisions. `no-ai-attribution` denies a commit, PR, or comment that carries AI authorship credit. `prose-question-guard` blocks a stop once to move a prose question into `AskUserQuestion`.

### Writing quality
- ap-style-check, flags AP Style violations on write/edit
- ai-slop-detector, warns about AI writing patterns
- accessibility-check, checks alt text, headings, link text
- copywriting-preflight, interviews for intent before drafting a new piece or revision

### Verification
- source-attribution-check, flags unattributed quotes or claims
- verification-reminder, prompts fact verification
- data-methodology-check, ensures methodology is documented

### Editorial workflow
- source-diversity-check, notes sourcing diversity
- legal-review-flag, flags potential defamation risk
- pre-publish-checklist, pre-publication reminder at session end
- deadline-tracker, surfaces upcoming deadlines at session start

### Preservation
- archive-reminder, reminds to archive URLs

### Interaction
- [prose-question-guard](https://skills.amditis.tech/prose-question-guard/), Stop hook that blocks once to move prose questions into AskUserQuestion; manual Python 3 install, Claude Code only

### Development
- one-way-door-check, blocks irreversible architectural decisions
- bug-report-detector, detects when user is describing a bug
- pre-commit-review, surfaces the staged diff for review before a commit and flags removed guardrails
- no-ai-attribution, blocks AI authorship credit in commits, PR bodies, and comments

## Plugins

Skills above are distributed as Claude Code plugins. Install them in Claude Code via:

```
/plugin marketplace add jamditis/claude-skills-journalism
/plugin install <plugin-name>@claude-skills-journalism
```

Restart your Claude Code session after installing or updating a plugin so the new skills and commands register.

Codex can use the repository's legacy-compatible marketplace route for the 16 nested `journalism-core` skills:

```bash
codex plugin marketplace add jamditis/claude-skills-journalism
codex plugin add journalism-core@claude-skills-journalism
```

This repository does not ship native Codex manifests. The route does not convert Claude-specific commands, agents, hooks, or root-level skills into Codex components.

Plugins (14): `journalism-core` (16 skills), `newsroom-agent-ops` (9), `research-toolkit` (6), `dev-toolkit` (14), `security-toolkit` (5 skills + /hotpatch command), `project-templates-toolkit` (4), `pdf-design` (1), `pdf-playground` (1), `web-design-picker` (1), `visual-explainer` (1), `video-toolkit` (4), `okf-wiki` (1), `superjawn` (13), `autocontext` (cross-session knowledge persistence, hooks, commands, and agents rather than skills).

## Alternate installation

Install the 16 `journalism-core` skills through the Agent Skills standard for Codex:

```bash
npx skills@latest add https://github.com/jamditis/claude-skills-journalism/tree/master/journalism-core --skill '*' --agent codex --copy -g -y
```

For Claude Code, clone the repo and copy or symlink individual skill directories into `~/.claude/skills/`:

```bash
git clone https://github.com/jamditis/claude-skills-journalism.git ~/projects/claude-skills-journalism
ln -sfn ~/projects/claude-skills-journalism/research-toolkit/skills/free-apis-catalog ~/.claude/skills/free-apis-catalog
```

## License

MIT. Use, adapt, and share freely.