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 skills add amazing-fish/agent-skills --skill report-for-humangit clone --depth 1 https://github.com/amazing-fish/agent-skillsWrote 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.
[](https://agentmods.dev/skills/amazing-fish/agent-skills/report-for-human)<a href="https://agentmods.dev/skills/amazing-fish/agent-skills/report-for-human"><img src="https://agentmods.dev/badge/skills/amazing-fish/agent-skills/report-for-human/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.
<a href="https://agentmods.dev/skills/amazing-fish/agent-skills/report-for-human"><img src="https://agentmods.dev/badge/skills/amazing-fish/agent-skills/report-for-human.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 5.1 | $0.00108 | $0.01744 |
| Opus 5.5 | $0.00043 | $0.00698 |
| Sonnet 5.5 | $0.00022 | $0.00349 |
| Haiku 4.5 | $0.00011 | $0.00174 |
Grade A, and why
report-for-human 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.
How it starts
The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.
给人看的汇报
由
explain-diff-for-human-review重构，保留GitHubxsy/agent-skills的来源与 Apache-2.0 归属；原上游版本与本地修改见仓库sources.lock.json。
让读者只看标题和主图，就能说出：现在是什么、为什么、影响谁、下一步做什么。完成深入取证，再把解释压缩成可视化 HTML。
先选要回答的问题
| 问题 | kind |
主视觉 | 取证重点 |
|---|---|---|---|
| 改了什么，是否值得接受 | change |
前后对照 + 影响 | 实际版本差异、调用方、验证和风险 |
| 要求到底是什么 | requirements |
要求 → 适用条件 → 验收 | 提出方、原文/契约、项目选择与推断 |
| 现在到了哪步，卡在哪里 | status |
阶段条 + 当前卡点 | 完成/进行/未知、依赖与最新证据时间 |
| 为什么这样，谁管哪段 | mechanism |
有标注箭头的流程/责任图 | 输入输出、分支条件、控制权与时序 |
- 按主问题选一个主视觉；必要时组合两个。小改动用紧凑前后对照，不铺完整架构；纯讲解不强造 PR 或合入结论。
- 创建新 PR 后立即给初版 HTML，标明仍在运行的 CI/Review；后续事实变化时更新同份报告，不等审查全绿才汇报。
- 汇报不授予实现、发送、发布或合并权限，不替用户批准；用户只要解释时，保持只读。
先把事实讲对
- 固定任务范围和取证时间，读取真实来源。区分事实、推断、未知、模拟；将证据缺口及其影响放在首屏可见处。注释和 README 不自动等于上游强制要求。
- 要求类报告区分用户要求、外部规范、接口契约、项目配置及建议；没有权威来源时写“待核实”，不要虚构标准编号或强制性。
- 现状类报告分开代码合入、上传受理、注册表核验、Release 和上游收录；计划触发、实际开始、完成与单条数据采集时间分别记录。同一时区展示。
- 改动类报告逐项对照需求及真实调用路径；PR 读取实际
base_sha/head_sha、差异、测试与审查状态，使用 PR 审查规则。非 PR 的本地差异保留快照标识。 - 本次执行、历史/CI、源码推断、未验证分别表述。跨轮证据复用、性能/交互、制品或特殊差异读取 证据边界。同树可复用适用源码证据，不能据此认定 tarball 字节相同。
复现这套审美
先读 视觉配方，查看 assets/reference-report.html；有图像工具时查看 assets/reference-preview.png。它们来自本次认可的 Piano 发布地图的视觉语言，内容已去项目化且明确为虚构示例。
默认使用随附生成器，继承版式而非重新凭印象设计：
python <skill-root>/scripts/render_report.py --input <verified-report.json> --output <report.html>
按 examples/change.json、requirements.json、status.json、mechanism.json 的同类结构填写已核验事实。生成器只负责呈现，不获取来源、不判断事实、不自动声称通过。先运行一份最接近的示例，再替换内容。
- 采用浅灰画布、白色大留白、深蓝结论区、蓝色重点和稳定的参与方配色。主标题约 32–44px、正文至少 16px；主视觉宽约 1180px，文字块保持短行。
- 首屏保留一个结果标题、一两句回答、一个当前状态或关键判断；随后一个主视觉、最多一两项下一步、一个折叠证据区。每张卡片表达一个关系，不把所有段落装成卡片。
- 生成器不能表达任务的关键分支时，沿用同一主题与组件扩展 HTML/SVG。读取 讲解图式，明确箭头传递什么、条件如何改变结果；不为装饰加图。
- 自定义时把
assets/report-theme.css与assets/report-layout.css内联到 HTML 的<style>；根节点使用.human-report，保留prefers-color-scheme: dark、打印、焦点与窄屏规则。不要从旧示例继承项目版本号、事实、标志或证据链接。
What ships with it
20 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- agents/openai.yaml 352 B
- assets/reference-preview.png 77 KB
- assets/reference-report.html 30 KB
- assets/report-layout.css 7.5 KB
- assets/report-theme.css 6.6 KB
- examples/change.json 1.3 KB
- examples/mechanism.json 5.2 KB
- examples/requirements.json 1.7 KB
- examples/status.json 1.7 KB
- references/design-sources.md 2.3 KB
- references/evaluation.md 2.3 KB
- references/evidence.md 7.3 KB
- references/explanation-patterns.md 6.0 KB
- references/independent-review-contract.md 2.8 KB
- references/pr-review.md 3.7 KB
- references/visual-design.md 3.0 KB
- scripts/classify_diff_size.py 3.5 KB runs code
- scripts/render_report.py 13 KB runs code
- scripts/resolve_report_path.py 2.4 KB runs code
- scripts/validate_independent_review.py 9.9 KB runs code
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 · 66 lines · 108 tokens per session scan A 30569da53fdf
report-for-human is a skill published in the GitHub repository amazing-fish/agent-skills (2 stars, last pushed 2d ago), licensed Apache-2.0. It adds 108 tokens to every session and 1,744 once invoked, about $0.0004 per session on Opus 5.5. 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 2026-10-22.
Other skills, from other repositories
daily-report-generator
Automatically generate daily construction reports from field data, worker inputs, weather, and progress photos. Creates professional PDF reports.
org-report
Generate an executive organization brief (leadership chain, direct reports, extended teams, and vendor partners) for a target person. Produces a Word + PDF report from a directory MCP source (WorkIQ is the reference implementation) and displays it in a Copilot canvas. Triggers on "generate org report for X", "build…
extract-straight-bill-of-lading-data
Extract structured fields from straight bill of lading documents.
to-checklist
A tool that turns unresolved decisions into a confirmation checklist PDF. Each item includes the current situation and possible choices so another person can decide without asking for the background again.
sprint-summary-pdf
Generate per-sprint PDF summaries with a card layout from docs/sprints. Use when the user asks for sprint summaries, sprint recap PDFs, shareable sprint story cards, or to drop PDFs into each sprint folder.
project-brief
Notebook-collage weekly digest template — spiral-bound reporter's notebook aesthetic with cobalt tape, mint folder tab, binder clip, receipt block, ruled baseline grid; built for AI-product / team / design weekly digests delivered as a downloadable high-res page.