A guided writing command that drafts text, checks it against your writing style, and removes signs of machine-written prose before giving you Markdown to edit and approve.

Use it to prepare articles, posts, emails, or other written pieces from a supplied brief. It helps compare drafts with your saved voice guide and clean them before final approval.

It keeps your own voice central and prevents text from being sent or rendered before you have reviewed it. It also avoids guessing at your style when the saved writing samples are too limited.

write — Command by robconery

Install

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.

agentmods

npx agentmods add commands/robconery/champion/write

Clone the repo

git clone --depth 1 https://github.com/robconery/champion

Per session 57 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 959 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What 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	$0.00057	$0.00959
Opus 5	$0.00028	$0.00479
Sonnet 5	$0.00011	$0.00192
Haiku 4.5	$0.00006	$0.00096

Measured yesterday against content hash a4ebf9d04701, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

write scanned grade A with 0 findings 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 yesterday.

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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

champion/commands/write.md · 45 lines

How it starts

The opening of the file, as written. The whole thing — 45 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Plugin note: this command ships in the champion plugin, so every council agent it names is a plugin agent. When spawning any agent mentioned below, use the champion: prefix for the agent type (e.g. champion:historian, champion:north-star, champion:viewer).

Setup gate: before doing anything else, check that ~/.champion/dossier/PROFILE.md exists. If it does not, stop immediately and tell the user, politely, that the council cannot work without knowing who it fights for: run /champion:setup first (a 20–30 minute interview). Do not proceed in any form until setup is complete.

write, the writing gauntlet

The piece:

$ARGUMENTS

This is a process, not a one-shot. The machine drafts and cleans. The user gets the last word. The center of gravity is their voice, and the iron rule is that nothing leaves in markdown form until they have edited and approved it.

Read first

~/.champion/dossier/VOICE.md: the rubric, taste, aesthetic, and their actual writing samples. The positive target. Write through it. If the samples are thin, say so and ask for more rather than guessing at a voice.
${CLAUDE_PLUGIN_ROOT}/references/voice-scrub.md: the cleaning checklist. The list of AI tells to strip.
For a newsletter or marketing email: ${CLAUDE_PLUGIN_ROOT}/references/marketing-playbook.md: teach a real outcome, link the lead magnet, invite softly. The newsletter is a flywheel asset, not an announcement.

The pipeline

Clarify the job. What is the piece, who reads it, what should they feel or do after. If it is not obvious from $ARGUMENTS, ask one tight question, then go.
Draft in voice (VOICE.md). Do not write "content." Write the way the user talks on the page: a real opinion, stated, then earned.
Voice-check. Read it whole against VOICE.md. Does it sound like the user or like a feed. Fix where it drifts.
Scrub against ${CLAUDE_PLUGIN_ROOT}/references/voice-scrub.md. Run the two sweeps that always catch something: the em-dash sweep and the punchy-fragment sweep. Then the rest of the ban list.

Read the full file on GitHub · 45 lines

Changes

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.

yesterday First seen · 45 lines · 57 tokens per session scan A a4ebf9d04701

Subscribe to this mod's changes

write is a command published in the GitHub repository robconery/champion (5 stars, last pushed 29d ago), licensed MIT. It adds 57 tokens to every session and 959 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Writing and docs

write

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

write, the writing gauntlet

Read first

The pipeline

What this file has done since we first saw it

Other commands, from other repositories

checklist

clarify

specify

analyze

converge

implement