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 dryvist/claude-code-plugins --skill fast-subagentgit clone --depth 1 https://github.com/dryvist/claude-code-pluginsWrote 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/dryvist/claude-code-plugins/fast-subagent)<a href="https://agentmods.dev/skills/dryvist/claude-code-plugins/fast-subagent"><img src="https://agentmods.dev/badge/skills/dryvist/claude-code-plugins/fast-subagent/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/dryvist/claude-code-plugins/fast-subagent"><img src="https://agentmods.dev/badge/skills/dryvist/claude-code-plugins/fast-subagent.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 | $0.00059 | $0.01146 |
| Opus 5.5 | $0.00024 | $0.00458 |
| Sonnet 5.5 | $0.00012 | $0.00229 |
| Haiku 4.5 | $0.00006 | $0.00115 |
Grade A, and why
fast-subagent scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
relative to this file; it needs only `bash`, `curl`, `jq`). It reads the How it starts
The opening of the file, as written. The whole thing — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.
fast-subagent
Doctrine: local-subagents (same plugin) says why and what to delegate and
how a busy single-slot tier behaves. This skill is the one-command how, tuned
for reuse: call it early, call it often, call it again.
1. When — every routine step, all session long
The Delegate column of local-subagents §1 is the list. The habit this skill
adds: do not decide once. Each time a new routine step appears — the fifth
log to reduce, the third file to summarize, the test output to read — send it
the same way. Repeated calls from one harness keep the backend's prompt cache
warm and the lock held, so the tenth call is cheaper than the first.
Keep for yourself: intent, architecture, tradeoffs, risk, anything you would not accept from a smaller model without checking. Never send secrets, secret store context, private topology, or personal data.
The router is not a Claude subagent and never writes code.
2. How — one command
printf '%s\n' "$INSTRUCTION" "$INPUT" | fast-subagent.sh
fast-subagent.sh lives in this skill's scripts/ directory (resolve it
relative to this file; it needs only bash, curl, jq). It reads the
endpoint from the environment — LLM_ROUTER_URL + LLM_ROUTER_TOKEN_FILE, or
the local proxy's OPENAI_API_BASE_URL + LITELLM_LOCAL_KEY — sends one
bounded request for role small, prints the answer on stdout and
model=<what answered> fallbacks=<n> on stderr.
Options: --model medium (a capability name), --system TEXT, --max-tokens N,
--timeout SEC, --raw (full JSON), --release (see §4).
Write the prompt for a small model — short imperatives, an explicit output shape, a row cap, a hard STOP:
Extract every failing test from the input. Do not explain or fix.
Output ONLY: test name | file | first error line. At most 40 rows. STOP.
Send only what the step needs. A call that carries your whole conversation costs more than doing the step yourself.
3. What the router decides, not you
What ships with it
1 file 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.
- 10d ago Changed · -17 lines e677cc7b1b3a
- 11d ago Changed · +17 lines f8368060ebe6
- 15d ago Changed · +2 lines 62d21f229bd2
- 1mo ago Changed · -57 tokens per session c01171859d01
- 1mo ago First seen · 100 lines · 116 tokens per session scan A e934ad234901
fast-subagent is a skill published in the GitHub repository dryvist/claude-code-plugins (3 stars, last pushed 5d ago), licensed Apache-2.0. It adds 59 tokens to every session and 1,146 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-09-19.
Other skills, from other repositories
career-portfolio
A portfolio-building skill for Korean job seekers that organizes projects and explains their work for development, design, marketing, and planning roles.
hr-operations
A skill for designing remote-work and team-operation policies. It covers flexible work, asynchronous communication, collaboration tools, productivity tracking, and employee experience.
collab-conflict
A guide for handling difficult workplace colleagues and recurring communication conflicts.
collab-feedback-loop
A guide for giving and receiving workplace feedback about specific actions and their effects.
collab-productivity-report
A Korean-language assistant for preparing weekly work reports. It organizes completed work, ongoing tasks, issues, and next-week plans into a set format.
book-publishing
Multi-book content, voice, and production workflows for nonfiction authoring. Use when planning a book, maintaining consistent voice across chapters, structuring a manuscript, or running an end-to-end publishing pipeline.