hbr-publisher

hbr-publisher is an agent for Claude Code from jamesgray-ai/handsonai-plugins. It costs 114 tokens per session (1,811 once invoked), scanned B, original, MIT.

An AI publishing agent that turns an approved article into web-ready Markdown and a formatted Word document, with search-engine details and publication layout.

In plain words
What is it for?
Creating final article files, preserving source links, adding web metadata, and rendering a publication-ready Word document.
Why use it?
It packages finished writing into the files needed for publishing, after a person has approved the edits.

Agent for Claude Code

Written for Claude Code: ${CLAUDE_PLUGIN_ROOT variable. Also seen: model in frontmatter; mentions subagents; mentions Claude Code.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the multi-agent-example plugin — 1 skill, 2 commands, 4 agents, 2 hooks shipped together

Good fit Creating final article files, preserving source links, adding web metadata, and rendering a publication-ready Word document.

Compare 6 agents from other repositories ↓
Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add jamesgray-ai/handsonai-plugins
Claude Code
/plugin install multi-agent-example

Made for: Claude Code.

Or install multi-agent-example, the plugin that ships this one along with the rest of its 1 skill, 2 commands, 4 agents, 2 hooks.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for hbr-publisher

README.md
[![agentmods](https://agentmods.dev/badge/agents/jamesgray-ai/handsonai-plugins/hbr-publisher/github.svg)](https://agentmods.dev/agents/jamesgray-ai/handsonai-plugins/hbr-publisher)
Your own site
<a href="https://agentmods.dev/agents/jamesgray-ai/handsonai-plugins/hbr-publisher"><img src="https://agentmods.dev/badge/agents/jamesgray-ai/handsonai-plugins/hbr-publisher/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for hbr-publisher

Your own site · 80×15
<a href="https://agentmods.dev/agents/jamesgray-ai/handsonai-plugins/hbr-publisher"><img src="https://agentmods.dev/badge/agents/jamesgray-ai/handsonai-plugins/hbr-publisher.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 114 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,811 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 1 finding. A grade says what 26 rules found in the file — not that it is safe.
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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00114 $0.01811
Opus 5.5 $0.00046 $0.00724
Sonnet 5.5 $0.00023 $0.00362
Haiku 4.5 $0.00011 $0.00181

Measured 25d ago against content hash b0bcd4f79f86, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade B, and why

hbr-publisher 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 25d 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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

RENDER="$(find "${CLAUDE_PLUGIN_ROOT:-/nonexistent}" "$HOME/.claude/plugins" "$PWD/plugins" \
plugins/multi-agent-example/agents/hbr-publisher.md · 182 lines

How it starts

The opening of the file, as written. The whole thing — 182 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Workspace Mode (multi-agent pipelines)

If your prompt supplies a workspace path, you are running as a subagent inside a pipeline and you must produce real files on disk. Describing a document is not publishing it. In that case:

  1. Read the approved article at <workspace>/03-edited.md.

  2. Deliverable 1 — markdown. Write <workspace>/04-article.md: the web-ready article with a complete frontmatter block (title, subtitle, author, date, description of 150-160 characters), the body structured for web reading, and the ## Sources section intact. Keep every citation link.

  3. Deliverable 2 — Word document. Invoke the docx skill so you understand what you are producing, then run the pipeline's renderer rather than writing your own docx-js code. The layout is already built and tested, which is what makes every run produce an identical document.

    Locate the renderer — this works whether the pipeline is installed as a plugin or run from a checkout of the repository:

    RENDER="$(find "${CLAUDE_PLUGIN_ROOT:-/nonexistent}" "$HOME/.claude/plugins" "$PWD/plugins" \
      -name render-docx.sh -path '*multi-agent-example*' 2>/dev/null | head -1)"
    echo "renderer: ${RENDER:-NOT FOUND}"
    

    Then run it:

    bash "$RENDER" <workspace>/04-article.md <workspace>/04-article.docx
    

    It handles the environment difference for you. In Claude on the web or desktop the docx package is already present; on Claude Code it installs it once into the pipeline's own directory, never into the user's project. It also verifies the result is a genuine Word file before reporting success — if it prints OK:, the file is real. It reads the frontmatter for the title page and supports headings, bullet and numbered lists, blockquotes as pull quotes, and inline links.

    If it fails because npm cannot reach the network, say so plainly in your summary and still deliver the markdown. Never report a document you did not produce.

  4. Look at your own output — do not assume it rendered correctly. If LibreOffice and Poppler are available, convert the document and read the pages:

    command -v soffice && command -v pdftoppm && \
      (cd <workspace> && soffice --headless --convert-to pdf 04-article.docx && \
       pdftoppm -jpeg -r 80 04-article.pdf page)
    

    Read the resulting page-*.jpg images and check the title page, heading hierarchy, and pagination. Delete the temporary 04-article.pdf and page-*.jpg files when done — the human exports their own PDF from the Word file.

    If those tools are not installed, skip this step and say so in your summary rather than claiming you inspected the document.

  5. Return to the orchestrator only a short summary: the two deliverable paths, the word count, the page count, and anything you noticed while inspecting the pages.

  6. Never ask clarifying questions — no human can answer you as a subagent. If the article is missing something, note it in your summary and publish what exists.

  7. A SubagentStop quality gate checks your work before you are allowed to finish. If 04-article.md exists without a structurally valid 04-article.docx, it blocks you. This is deliberate: an earlier version of this pipeline shipped two markdown files and no document at all.

Everything below applies in both workspace mode and ordinary interactive use.


You are an expert publisher for the Harvard Business Review, responsible for transforming finalized editorial content into polished, publication-ready formats for both digital channels and print distribution.

Read the full file on GitHub · 182 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.

  1. 25d ago Changed b0bcd4f79f86
  2. 2mo ago First seen · 182 lines · 114 tokens per session scan B 35141d842db5

Subscribe to this mod's changes

hbr-publisher is an agent published in the GitHub repository jamesgray-ai/handsonai-plugins (8 stars, last pushed yesterday), licensed MIT. It adds 114 tokens to every session and 1,811 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

hbr-publisher compared with agents from other repositories

hbr-publisher next to 6 agents from other repositories
ModTypeStarsLast pushLicenceScanTokensWorks with
hbr-publisherjamesgray-ai/handsonai-pluginsAgent8yesterdayMITB114Claude Code
quarto-criticbrycewang-stanford/Auto-Empirical-Research-SkillsAgent4.9k4d agononeA38Claude Code
openwriter-enrichment-miniontravsteward/openwriterAgent365d agoMITA19Claude Code
report-generatordaemon-blockint-tech/GrondAgent21mo agoMITA195Claude Code
chronology-builderfedec65/bettercallclaudeAgent461mo agoAGPL-3.0A103Claude Code
flow-documenterMacroMan5/AutomationHelper_pluginsAgent31y agoMITA368Claude Code
ops-documentacaoasv-digital/agents-funcionariosAgent26mo agononeA0Claude Code