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/rashadansari/myagents/plugin-authoringnpx skills add RashadAnsari/myagents --skill plugin-authoringgit clone --depth 1 https://github.com/RashadAnsari/myagentsWrote 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/rashadansari/myagents/plugin-authoring)<a href="https://agentmods.dev/skills/rashadansari/myagents/plugin-authoring"><img src="https://agentmods.dev/badge/skills/rashadansari/myagents/plugin-authoring.svg" alt="Measured on agentmods" 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 | $0.00081 | $0.01066 |
| Opus 5 | $0.00041 | $0.00533 |
| Sonnet 5 | $0.00016 | $0.00213 |
| Haiku 4.5 | $0.00008 | $0.00107 |
Grade A, and why
plugin-authoring 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 4d 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 — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Plugin Authoring
Rules for writing skills, commands, and agents. Apply them when creating or editing any of these files.
Universal Rules
- Assume the model is already smart. Only write what it does not already know: project facts, exact procedures, output contracts. Cut explanations of general concepts. Challenge every line: "does this justify its token cost?"
- Write descriptions in third person, stating both what the artifact does and when to use it. Include the trigger phrases a user would naturally say. Put the key use case first: listings truncate long descriptions.
- Use one consistent term per concept throughout a file. Do not mix synonyms.
- No time-sensitive content. Nothing that becomes wrong on a date.
- Forward slashes in all paths.
- Platform-neutral wording: the same file must read correctly in every editor that loads it. No editor-specific product names or jargon in prose.
- Never reference another plugin file by repo-relative path. The install location differs from the source repo. Invoke skills by name and locate scripts via the plugin or skill root variable.
Skills
Frontmatter: name (lowercase letters, numbers, hyphens, max 64 chars) and description (max 1024 chars, non-empty). Everything else is optional.
- Keep the body under 500 lines. When approaching the limit, split detail into supporting files in the skill directory and link them from SKILL.md, one level deep only. Give reference files longer than 100 lines a table of contents.
- Match specificity to fragility. Judgment tasks get short, high-freedom guidance. Fragile sequences get exact commands with "do not modify" instructions. One default approach with an escape hatch beats a menu of options.
- For multi-step workflows, write numbered steps. For quality-critical output, build a feedback loop: produce, validate, fix, repeat until the validator passes.
- Side-effect workflows the user must trigger themselves (commit, deploy, publish) would normally set
disable-model-invocation: true, but this repo omits the flag for now: some editors currently hide plugin-delivered skills with it from the command menu entirely, making them un-invokable. Re-add it once that bug is fixed. Background knowledge that is not a meaningful action: setuser-invocable: false. - Bundled scripts: instruct execution, not reading, unless the algorithm itself is the reference. Scripts must handle their own error cases instead of failing back to the model, and every constant needs a justifying comment.
- A loaded skill stays in context for the whole session. Write standing instructions, not one-time steps.
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.
- 4d ago First seen · 58 lines · 81 tokens per session scan A 3d8cbb772675
plugin-authoring is a skill published in the GitHub repository RashadAnsari/myagents (6 stars, last pushed 27d ago), licensed MIT. It adds 81 tokens to every session and 1,066 once invoked, about $0.0004 per session on Opus 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 2026-08-31.
Other skills, from other repositories
apple-music
Apple Music integration — AppleScript (local Music.app) and the Apple Music API / web player (cross-platform library, playlists, playback, and queue).
outlook
Use Microsoft Outlook end-to-end — read, search, triage, and send mail; manage calendar events and meeting invites; work with contacts, tasks, color categories, and mail rules; check Out-of-Office. Trigger this skill whenever the user asks to do anything with their inbox, email, calendar, meetings, contacts, or to-do…
vibebrowser
Control the user's real local browser through the Vibe Browser CLI bridge. Use for ANY task against the user's own Vibe-connected browser — their tabs, cookies, logged-in sessions, extensions, opening pages, snapshots, clicks. ALWAYS load this skill before responding to a browser request, so you can recover the user's…
setup
Set up Vibe Browser so Claude Code can drive the user's real Chrome. Use when the vibe MCP tools are missing, return "No connection to Vibe extension", time out, or when the user asks how to install/connect/configure Vibe, the Vibe Chrome extension, or remote browser control.
create-pr
Automated end-to-end workflow to verify, format, test, push, and open a polished, AI-native Pull Request using GitHub CLI (gh) with Conventional Commits and rich markdown formatting.
gsc-cannibalization-check
Find queries where several pages of the same site compete against each other in Google, and decide which page should win. Use when the user mentions cannibalisation, duplicate or overlapping pages, pages competing for the same keyword, or asks why a page will not rank.