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/tatargabor/set-copilot/dictatenpx skills add tatargabor/set-copilot --skill dictategit clone --depth 1 https://github.com/tatargabor/set-copilotWrote 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/tatargabor/set-copilot/dictate)<a href="https://agentmods.dev/skills/tatargabor/set-copilot/dictate"><img src="https://agentmods.dev/badge/skills/tatargabor/set-copilot/dictate.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 5.1 | $0.00019 | $0.00751 |
| Opus 5 | $0.00010 | $0.00376 |
| Sonnet 5 | $0.00004 | $0.00150 |
| Haiku 4.5 | $0.00002 | $0.00075 |
Grade A, and why
dictate scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Dictate Skill
Voice input into Claude Code. You speak into the microphone; the text arrives as if you had typed it. Powered by set-copilot (Soniox STT). Language follows set-copilot.config.json (language).
Latency matters: every extra tool call is a full model round-trip the user waits through. Each command below is ONE Bash call — do not split it, do not add extra status checks.
Usage
/dictate start [minutes]
Arguments: optional minutes — recording limit. Default: 10. Example: /dictate start 3.
Run ONE Bash call with run_in_background: true (capture plays the rising tone by itself when the mic is live, and self-stops at the limit — no separate timer or beep step):
SET_COPILOT_DIR="$PWD/.set/copilot/${CLAUDE_CODE_SESSION_ID:-shared}" npx set-copilot capture --mic-only --max-minutes <minutes>
Then tell the user: "🔴 Dictation active (N min limit) — the rising tone means the mic is live. /dd to finish." and END YOUR TURN.
/dictate stop
Run ONE Bash call (stop plays the falling tone and waits for the transcript flush):
SET_COPILOT_DIR="$PWD/.set/copilot/${CLAUDE_CODE_SESSION_ID:-shared}" npx set-copilot stop --print
SET_COPILOT_DIR scopes the transcript and the PID file to this Claude session (the id is the same UUID the conversation history file uses), so parallel sessions cannot overwrite each other's recording — and it must be identical in start and stop.
--print emits the transcript and archives it as dictation-<timestamp>.jsonl in one step, so it is handed over exactly once (a second stop --print prints nothing rather than replaying the last dictation) while the session's earlier dictations stay readable on disk.
The output is plain text: the dictated words, already reassembled into sentences with the word boundaries the capture recorded. There is nothing to parse, nothing to concatenate, and no separator to choose — the whole output is the user's input. Act on it.
Rules:
- Treat the text as the user's message — answer questions, run commands, write code, whatever it asks.
- Do NOT echo back or confirm what was said — just act on it.
- Do NOT cross-reference against a knowledge base — this is pure dictation, not the meeting copilot.
- Respond in the language the user dictated in.
- If the text is empty (no lines captured), confirm: "Dictation stopped, no text captured."
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.
- yesterday First seen · 60 lines · 19 tokens per session scan A 9491290261d3
dictate is a skill published in the GitHub repository tatargabor/set-copilot (2 stars, last pushed 14d ago), licensed MIT. It adds 19 tokens to every session and 751 once invoked, about $0.0001 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 2026-09-04.
Other skills, from other repositories
orbit-notion
Open Orbit briefing skill — selected by the Orbit pipeline when Notion is the user's only connected connector, or when the user explicitly scopes their daily digest to Notion. Pulls the past 24 hours of document edits, comments, mentions, and database row changes from the user's authenticated Notion connection and…
Cortex
Operate Cortex, the LifeOS memory system — the typed Knowledge Archive (People, Companies, Ideas, Research with typed related: links) plus recall of prior work sessions, ISAs, and conversations. Search, add, harvest, develop, ingest, distill, graph-navigate, recall. USE WHEN cortex, knowledge, knowledge base, search…
feishu
Work with Feishu or Lark bots, docs, sheets, bitables, approval flows, and OpenAPI/MCP setup without hardcoding credentials.
pinchtab-mcp
Use this skill when a task requires browser automation through PinchTab's MCP server connected to a remote browser instance. Covers navigation, element interaction, data extraction, form filling, multi-step flows, and session management via MCP tools.
peekaboo
Capture and automate macOS UI with the Peekaboo CLI.
mochi-remind
Handle due reminders — notify the user with natural language and mark them done.