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 ondrej-svec/heart-of-gold-toolkit --skill copy-editorgit clone --depth 1 https://github.com/ondrej-svec/heart-of-gold-toolkitWrote 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/ondrej-svec/heart-of-gold-toolkit/copy-editor)<a href="https://agentmods.dev/skills/ondrej-svec/heart-of-gold-toolkit/copy-editor"><img src="https://agentmods.dev/badge/skills/ondrej-svec/heart-of-gold-toolkit/copy-editor/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/ondrej-svec/heart-of-gold-toolkit/copy-editor"><img src="https://agentmods.dev/badge/skills/ondrej-svec/heart-of-gold-toolkit/copy-editor.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high System Prompt Leakage · line 83 Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.Fix: Remove any instructions that reveal, print, or output system prompts or internal rules. System instructions should never be exposed to end users.
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 | $0.00202 | $0.03938 |
| Opus 5.5 | $0.00081 | $0.01575 |
| Sonnet 5.5 | $0.00040 | $0.00788 |
| Haiku 4.5 | $0.00020 | $0.00394 |
Grade A, and why
copy-editor 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 2mo 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 — 301 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Copy Editor
I review copy. Two layers: the mechanical one that a script closes, and the judgment one that I surface but never close. A human always owns the final seal on voice, clarity, and readability.
This skill is agent-agnostic: any agent with Bash and Read can run it. The script is the portable contract.
Boundaries
This skill MAY: read files under the configured scope, run the copy-audit script, parse its JSON output, read the repo-local style guide and reject list, propose rewrites with rationale, draft a review note, report findings.
This skill MAY NOT: edit source files directly, mark Layer 2 judgment work as "approved", skip the deterministic Layer 1 pass, invent rules not present in the loaded language profile or repo-local rule files, close any gate the verification boundary assigns to a human.
Verification Boundary
Layer 1 (deterministic) is the only layer this skill may auto-close. Layer 2 (judgment) always produces suggestions, never verdicts. This skill will not mark a file or scope as "copy-editor approved" at Layer 2. That seal belongs to a human reviewer, whose signoff is recorded in the review note this skill drafts.
This clause is non-negotiable. If a repo's doctrine relaxes it, stop and ask before proceeding.
Common Rationalizations
| Shortcut | Why It Fails | The Cost |
|---|---|---|
| "Skip Layer 1, it's just formatting" | Layer 1 is the mechanical gate that lets Layer 2 focus on judgment. Skipping produces noisy Layer 2 reviews where the reviewer keeps finding typography issues instead of voice ones. | Review fatigue; human reviewer spends their attention on the wrong layer. |
| "Mark the scope approved at Layer 2 because nothing looks wrong" | Layer 2 is judgment. The skill does not have the cultural or situational context a human reviewer has. | False confidence; Czech voice drift that the skill cannot detect. |
| "Apply clarity rules to presenter scenes too" | Presenter scenes are outline hints a facilitator explains live. Applying participant-strict rules to them produces noise and misses the point. | Lots of false positives on valid outline-style content; suggestions get ignored. |
| "Invent rules the profile does not contain" | The profile is the contract. Inventing rules breaks reproducibility across runs. | Suggestions vary run to run; the skill becomes unreliable. |
What ships with it
16 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.
- knowledge/config-schema.md 8.1 KB
- knowledge/language-profiles.md 6.0 KB
- knowledge/lockfile-schema.md 12 KB
- knowledge/output-contract.md 5.7 KB
- knowledge/ROLE.md 14 KB
- knowledge/segmentation-prompt-v1.md 24 KB
- knowledge/segmentation-prompt-v2.md 19 KB
- knowledge/segmentation.md 11 KB
- rules/czech.ts 17 KB runs code
- rules/english.ts 2.1 KB runs code
- rules/index.ts 901 B runs code
- rules/types.ts 4.2 KB runs code
- scripts/copy-audit.ts 48 KB runs code
- scripts/finding-filter.ts 1.9 KB runs code
- scripts/lockfile.ts 12 KB runs code
- scripts/self-test.ts 24 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.
- 2mo ago First seen · 301 lines · 202 tokens per session scan A e61fcb2e6ffd
copy-editor is a skill published in the GitHub repository ondrej-svec/heart-of-gold-toolkit (19 stars, last pushed 1mo ago), licensed MIT. It adds 202 tokens to every session and 3,938 once invoked, about $0.0008 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 2026-08-30.
copy-editor compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| copy-editorondrej-svec/heart-of-gold-toolkit | Skill | 19 | 1mo ago | MIT | A | 202 | Claude Code |
| imagegenopenai/codex | Skill | 134k | today | Apache-2.0 | A | 113 | Codex |
| playwright-climicrosoft/playwright | Skill | 99k | today | Apache-2.0 | A | 19 | Claude Code |
| local-ai-agentsmicrosoft/ai-agents-for-beginners | Skill | 79k | today | MIT | A | 200 | Claude Code, Codex |
| systematic-debuggingobra/superpowers | Skill | 314k | 3d ago | MIT | A | 21 | Claude Code |
| next-cache-components-adoptionvercel/next.js | Skill | 143k | yesterday | MIT | A | 95 | Claude Code |
| next-partial-prefetching-adoptionvercel/next.js | Skill | 143k | yesterday | MIT | A | 103 | Claude Code |