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 anton-abyzov/vskill --skill html-reportgit clone --depth 1 https://github.com/anton-abyzov/vskillWrote 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/anton-abyzov/vskill/html-report)<a href="https://agentmods.dev/skills/anton-abyzov/vskill/html-report"><img src="https://agentmods.dev/badge/skills/anton-abyzov/vskill/html-report/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/anton-abyzov/vskill/html-report"><img src="https://agentmods.dev/badge/skills/anton-abyzov/vskill/html-report.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 | $0.00218 | $0.01833 |
| Opus 5.5 | $0.00087 | $0.00733 |
| Sonnet 5.5 | $0.00044 | $0.00367 |
| Haiku 4.5 | $0.00022 | $0.00183 |
Grade A, and why
html-report 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 6d 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.
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 — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.
html-report: reports that read as documents
A report is read, not parsed. A wall of prose with inline code chips is hard to scan; the same facts as a document (properties first, a title, one lead paragraph, short sections) read in seconds. This skill makes that the default for every model and host.
How the good-looking hosts do it
Claude desktop's polished document look is not an HTML artifact. It is a markdown file that the app renders: flat YAML frontmatter becomes a property table (list values become chips), the # H1 is a serif title, the first paragraph is the lead, ## sections follow. HTML artifacts are kept for interactive or visual content. SpecWithCode renders worker reports the same way, and Obsidian shows the same frontmatter as properties.
So the optimal path is:
- Default: structured markdown. Cheap, works in every model, every host renders it well, and it degrades to clean text anywhere else.
- Visual evidence: a self-contained HTML page built from
template.html, only when screenshots, a comparison, a results matrix or a chart say more than prose.
1. Structured markdown (default)
---
title: Show harness on thread rows
status: done
branch: feat/thread-harness-label
tests: 262/262 pass
blockers: none
tags: [web, sidebar, models]
---
# Show harness on thread rows
Thread rows now show the harness first, then the model: "Claude Code · Opus 5.5", "Copilot · Claude Opus 5.5". Before, native threads showed just "Claude".
## Evidence
- Commit `d7c074f0b` on `feat/thread-harness-label`, not pushed.
- Headless screenshots: `threads-light.png`, `threads-dark.png`.
- The 5 full-suite failures also fail on clean `origin/main`.
## Next
- Push and open the PR once the release branch is cut.
Rules (these match what document viewers accept):
- Frontmatter is flat:
key: valuelines, and lists as[a, b]or- alines. No nesting, no|/>block text, no duplicate keys; any of those makes viewers drop the whole table. - Keys are lowercase words (
review_cycleshows as "Review cycle"). Put short facts here: status, branch, version, test totals, blockers, owners, dates. Lists become chips, so use them for tags, people, files. - One
# H1, then one lead paragraph of one to three sentences: the outcome. Then##sections with bullets, one fact per bullet. - Use
> [!NOTE],> [!WARNING]or> [!CAUTION]callouts for the one thing the reader must not miss. - Tables only for real comparisons; numeric columns right-align on their own.
- No raw HTML in markdown; viewers strip it.
- Call the desktop app SpecWithCode on first mention and "the app" after that, never "Studio" or "SpecWeave Studio". Branch, repo, path and tool names stay literal (
feat/studio-plan-progress,studio_thread_list).
What ships with it
2 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.
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.
- 6d ago Changed · +28 tokens per session 3cc9fb0d614d
- 7d ago First seen · 110 lines · 190 tokens per session scan A f441ac531ff3
html-report is a skill published in the GitHub repository anton-abyzov/vskill (48 stars, last pushed yesterday), licensed MIT. It adds 218 tokens to every session and 1,833 once invoked, about $0.0009 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 2026-10-16.
Other skills, from other repositories
start-coding-task
Start a CodeAgent coding task from the user's request on one of their existing sessions, and report the session id so they can follow it.
review-agent-work
Review the changes a CodeAgent session made, and open or review a pull request when the user asks.
continue-with-codeagent-credits
Switch a session to a CodeAgent managed agent billed from the user's prepaid credits, with the user's explicit confirmation.
supervise-session
Monitor, redirect, answer or stop an existing CodeAgent session.
prereview
Review the current diff against this team's conventions and invariants before it ships. Invoke with /prereview before opening a PR, or run as a pre-push gate.
agent-config-setup
Generate AI agent config files for any project — Cursor, Claude Code, Copilot, Windsurf, Cline, AGENTS.md. Use when setting up a new project for AI agents.