Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add V-Songbird/hush/plugin install hushWrote 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/v-songbird/hush/pick-style)<a href="https://agentmods.dev/skills/v-songbird/hush/pick-style"><img src="https://agentmods.dev/badge/skills/v-songbird/hush/pick-style/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/v-songbird/hush/pick-style"><img src="https://agentmods.dev/badge/skills/v-songbird/hush/pick-style.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 | $0.00082 | $0.01040 |
| Opus 5 | $0.00041 | $0.00520 |
| Sonnet 5 | $0.00016 | $0.00208 |
| Haiku 4.5 | $0.00008 | $0.00104 |
Grade A, and why
pick-style scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 62 lines — stays where its author put it; the contents beside it link to each section on GitHub.
hush:pick-style
Picks which output style hush delivers, and delivers it. Every style listed here carries hush's mechanics — the silence between tool calls, the one structured final message, the hard caps. They differ only in voice and in what the final message is built to do.
Everything below is mechanical: two scripts do the reading, matching, and file-writing. Run their output verbatim — no re-parsing frontmatter by hand, no free-form judgment about which entry the user means.
1. List the shelf and ask for a number
Run:
node "${CLAUDE_PLUGIN_ROOT}/scripts/list-styles.js"
It prints one JSON object: styles (each with index, name, description, source, path, active), plus activeName, activeOnShelf, stockBackupExists, and restoredOverTakeover.
Render it as exactly this table, one row per entry in styles, in index order:
| # | Name | Description | Source | Active |
| --- | --- | --- | --- | --- |
| 1 | <name> | <description> | <source> | ✓ (only on the active row) |
source is stock for the benchmarked default and crafted for the user's own variants. Print it as it comes — it is what tells two same-named entries apart.
If restoredOverTakeover is true, add one line above the table: "A plugin update restored stock Hush over a prior takeover." Otherwise add nothing.
If activeOnShelf is false, add one line above the table instead: "<activeName> is active but its file is no longer on disk — no row below is checked." This means no row in the table will show ✓.
If the invocation already named a style, match it case-insensitively against styles[].name. Exactly one match → skip straight to step 2 with that entry's path, no table shown. No match, or more than one → show the table as above and ask the user to reply with a row number.
End the message with the table (or the skipped-straight-through report) — nothing after it. Wait for the reply.
2. Activate
Take the number from the user's reply and look it up in the styles array from step 1 — a plain index lookup, not a re-read of any file. An out-of-range or non-numeric reply gets the table shown again, unchanged, with no other action taken.
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 First seen · 62 lines · 82 tokens per session scan A 79a10359cbe5
pick-style is a skill published in the GitHub repository V-Songbird/hush (49 stars, last pushed 4d ago), licensed MIT. It adds 82 tokens to every session and 1,040 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 2026-08-30.
Other skills, from other repositories
knowledge-base
Keep verified findings in memory/knowledge/ so the same research is never done twice. Use before researching any topic, when a finding passes the verify gate, or when about to answer from something learned in an earlier session.
checkpoint
Maintain a thirty-second return point (CHECKPOINT.md) and a single-use handoff (HANDOFF.md) so any session can resume mid-goal without archaeology. Use when a large unit of work finishes, a session enters its closing stretch, when leaving instructions for the next session, or when resuming and the user asks where…
verify-gate
Treat research results and model knowledge as drafts until verified. Use when researching, quoting numbers or sources, recording knowledge as fact, or before declaring a task successful.
s-compact
Session handoff for the next session — write-side pair of /s-continue. Triggers on "s-compact", "session handoff", "handoff prompt", "hand off", "wrap up the session", "prep for next session". Distills what /s-continue cannot recover: subagent findings, tool-output numbers, and decisions that never entered the…
goal
Run a large or unfamiliar goal through the full ballast pipeline — mobilize what you already hold, terrain scan, full skeleton, atomic foundation learning with verification, then build from bedrock to a verified done. Use when the user hands over a big goal, enters a new field, or asks to learn X in order to achieve Y.
proof-standard
Never make an external-facing claim about a product without evidence from a truth file. Use when writing marketing copy, announcements, docs, landing pages, investor material, or answering "can our product do X".