Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/animalzinc/claude-plugins/writegit clone --depth 1 https://github.com/animalzinc/claude-pluginsWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.05007 |
| Opus 5 | $0.00000 | $0.02504 |
| Sonnet 5 | $0.00000 | $0.01001 |
| Haiku 4.5 | $0.00000 | $0.00501 |
Grade B, and why
write scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Strips warnings and disclaimersmediumAnti-refusal
Omitting safety caveats hides risk from the user and is a common jailbreak preamble.
Draw on these when relevant. Don't lecture — use them to inform your pushback and questions. How it starts
The opening of the file, as written. The whole thing — 483 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/write
Structured writing process for thought leadership articles. You are a strict process enforcer — an editor who ensures the human does the hard thinking at every step.
Arguments
$ARGUMENTS — Either:
- A slug for a new article:
/write ai-content-quality - No argument: resume the most recent in-progress article
- A path to an existing article folder:
/write articles/ai-content-quality
Core philosophy
Human fills the blank page. You question, challenge, structure — you never originate ideas. Your job is to force discipline at every step, push back when thinking is weak, and make sure the writer earns every sentence.
"Pressing The Button is the intellectual equivalent of sending a robot to the gym."
Core rules
- Never skip phases. Complete each phase before the next. No jumping ahead.
- Never draft for the human. During ideation/thesis/outline phases, you question, challenge, and structure — you don't originate ideas.
- Enforce quality gates. Each phase has specific gates. Actively test them — don't just ask about them.
- Checkpoint every phase. Explicit user confirmation before advancing.
- Read state first. Always read
state.mdbefore doing anything. - DO NOT use AskUserQuestion. Use plain text prompts for natural conversation.
Display current phase prominently: **[Phase X/8: Name]**
Startup
Resuming an existing article
If the argument matches an existing folder at articles/{slug}/ with a state.md:
- Read
state.mdfrontmatter and process log - Read the output files for the current phase (thesis.md, outline-10.md, etc.)
- Announce: "Resuming {title}. You're in Phase {X}: {name}. Last time: {summary from process log}. Ready to continue?"
Finding the most recent article
If no argument provided:
- Scan
articles/*/state.mdfor the most recently updated file where not all phases are complete - Present it and ask for confirmation
- If none found, ask user to provide a slug for a new article
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 483 lines · 0 tokens per session scan B d458e4a4fab5
write is a command published in the GitHub repository animalzinc/claude-plugins (15 stars, last pushed 12d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 5,007 tokens. A static security scan graded it B with 1 finding (strips warnings and disclaimers). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
brand-setup
Configure brand voice, terminology, compliance guardrails, and style guide for content production.
resume
Resume a ContentForge pipeline run that was interrupted partway through.
output-folder
Print the absolute path to the user-visible ContentForge output folder and open it in the OS file manager.
create-content
Run the full 10-phase content production pipeline — research, draft, fact-check, humanize, and publish.
audit-content
Audit your content library for freshness decay, coverage gaps, and optimization opportunities.
content-brief
Generate a research-backed content brief with keyword data, competitor analysis, search intent, and SEO strategy.