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 skills/millionco/debug-agent/skillnpx skills add millionco/debug-agent --skill skillgit clone --depth 1 https://github.com/millionco/debug-agentWhat 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 | $0.00060 | $0.03018 |
| Opus 5 | $0.00030 | $0.01509 |
| Sonnet 5 | $0.00012 | $0.00604 |
| Haiku 4.5 | $0.00006 | $0.00302 |
Grade A, and why
debug-agent scanned grade A 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 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **If reproduction is straightforward** (e.g., a single CLI command, a curl request, a simple script): write and run an ad hoc reproduction script yourself. Tailor it to the runtime — Playwright/Puppeteer for browser bu How it starts
The opening of the file, as written. The whole thing — 233 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Debug Mode
You are now in DEBUG MODE. You must debug with runtime evidence.
Why this approach: Traditional AI agents jump to fixes claiming 100% confidence, but fail due to lacking runtime information. They guess based on code alone. You cannot and must NOT fix bugs this way — you need actual runtime data.
Your systematic workflow:
- Generate 3-5 precise hypotheses about WHY the bug occurs (be detailed, aim for MORE not fewer)
- Instrument code with logs (see Logging section) to test all hypotheses in parallel
- Reproduce the bug.
- If a failing test already exists: run it directly.
- If reproduction is straightforward (e.g., a single CLI command, a curl request, a simple script): write and run an ad hoc reproduction script yourself. Tailor it to the runtime — Playwright/Puppeteer for browser bugs, a Node/Python/shell script for backend bugs, etc.
- Otherwise: ask the user to reproduce it. Provide clear, numbered steps. Remind them to restart apps/services if instrumented files are cached or bundled. Offer: "If you'd like me to write a reproduction script instead, let me know."
- Once the user confirms a reproduction pathway (manual or automated), reuse it for all subsequent iterations without re-asking.
- Analyze logs: evaluate each hypothesis (CONFIRMED/REJECTED/INCONCLUSIVE) with cited log line evidence
- Fix only with 100% confidence and log proof; do NOT remove instrumentation yet
- Verify with logs: ask user to run again, compare before/after logs with cited entries
- If logs prove success and user confirms: remove all instrumentation by searching for
#region debug log/#endregionmarkers and deleting those blocks (see Cleanup section). If failed: FIRST remove any code changes from rejected hypotheses (keep only instrumentation and proven fixes), THEN generate NEW hypotheses from different subsystems and add more instrumentation - After confirmed success: explain the problem and provide a concise summary of the fix (1-2 lines)
Critical constraints:
- NEVER fix without runtime evidence first
- ALWAYS rely on runtime information + code (never code alone)
- Do NOT remove instrumentation before post-fix verification logs prove success and user confirms that there are no more issues
- Fixes often fail; iteration is expected and preferred. Taking longer with more data yields better, more precise fixes
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 · 233 lines · 60 tokens per session scan A 86cdbb0fea28
debug-agent is a skill published in the GitHub repository millionco/debug-agent (302 stars, last pushed 3mo ago), licensed MIT. It adds 60 tokens to every session and 3,018 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
edulab
面向中高考数学的可视化解题工具：空间几何题生成可交互 3D 解题演示（建系向量法 + sympy 校验 + Three.js），函数表达式生成参数可控的 2D 图形（滑块实时联动）。触发词 /edulab。.
agent-browser
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a…
perplexity-webui-search
Use when the user asks to search Perplexity, query Perplexity through the web UI, perform web research with Perplexity, save a Perplexity answer as Markdown, or retrieve a Perplexity answer through browser automation.
social-push
使用 agent-browser 帮用户将内容发到社交媒体上。当用户需要发布内容、推送文章、上传文章、发帖到社交平台时使用此 skill。.
taoguba-crawler
This skill should be used when the user asks to "crawl taoguba", "crawl tgb", "scrape taoguba articles", "run the crawler", "crawl bbs", "crawl home page", "generate article HTML", or needs to run the Taoguba (tgb.cn) web crawlers.
dashi-ppt
制作 PPT、演示文稿、幻灯片、汇报材料时使用。Dashi PPT 基于预置视觉主题组合页面,生成可离线打开、可在浏览器编辑的 HTML 演示,支持导出 PPTX / PDF 文件。.