What is voice-refresh?

A command that updates a founder's writing-style reference from recent Granola meeting transcripts. Granola is a meeting-recording and transcription tool, and the process keeps only the founder's identified speech.

Use it to gather meetings since the last refresh, extract the founder's speech, check for transcription problems, run the update process, and present changes for approval.

It makes voice updates traceable and prevents unreviewed or poorly diarized meeting content from being added automatically.

voice-refresh — Command for Claude Code by assafkip/kipi-system

Install

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.

Clone the repo

git clone --depth 1 https://github.com/assafkip/kipi-system

Made for: Claude Code.

Or install kipi-core, the plugin that ships this one along with the rest of its 10 skills, 9 commands.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/commands/assafkip/kipi-system/voice-refresh/github.svg)](https://agentmods.dev/commands/assafkip/kipi-system/voice-refresh)

Your own site

<a href="https://agentmods.dev/commands/assafkip/kipi-system/voice-refresh"><img src="https://agentmods.dev/badge/commands/assafkip/kipi-system/voice-refresh/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.

Your own site · 80×15

<a href="https://agentmods.dev/commands/assafkip/kipi-system/voice-refresh"><img src="https://agentmods.dev/badge/commands/assafkip/kipi-system/voice-refresh.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 59 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 960 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

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	$0.00960
Opus 5.5	$0.00024	$0.00384
Sonnet 5.5	$0.00012	$0.00192
Haiku 4.5	$0.00006	$0.00096

Measured 7d ago against content hash 9b044f585b43, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

voice-refresh 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 7d 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.

plugins/kipi-core/commands/voice-refresh.md · 70 lines

How it starts

The opening of the file, as written. The whole thing — 70 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/voice-refresh

Interactive monthly refresh of the founder-voice skill. This command runs in a session because Stage 1 needs the Granola MCP, which is not reachable headless. PRD prd-voice-refresh-monthly-2026-07-04.

Steps

Determine lookback. Read automation/.voice-refresh-last (an ISO-8601 UTC instant). Lookback = since that timestamp; if missing, last 30 days. An explicit since <date> argument overrides. The orchestrator in step 4 writes that file itself, so do not write it here and do not treat its absence as normal: a missing stamp after a completed run is a defect, not a first run.
Pull (Stage 1, MCP). List Granola meetings in the window and pull each transcript. Write the usable ones to the corpus input dir. Then run: python3 q-system/.q-system/scripts/granola-voice-harvest.py <input_dir> q-system/output/voice-corpus The harvest isolates Me: speech and flags any degraded-diarization meeting.
Accrete, do not replace. Keep meetings from the last 12 months; age out older ones. New meetings append to the corpus input dir.
Orchestrate (Stages 2-3, headless-safe). Run: python3 automation/voice_refresh.py q-system/output/voice-corpus It REFUSES to proceed if any meeting is review-flagged (contamination gate), and STOPS with an environmental-trigger diagnosis if claude is unavailable. It emits q-system/output/voice-corpus/voice-delta.md, stamps automation/.voice-refresh-last, and writes automation/.voice-delta-pending when the delta has a personal half.
Ship the two halves separately. voice-delta.md is split:
- Routine. Style patterns. Merge them through a reviewed PR without waiting on the founder: edit the skeleton voice-dna.md, bump the kipi-core plugin.json version, commit, open the PR, and let review run. Never push them straight to main.
- Personal. Patterns mined from a personal disclosure (layoffs, a mental-health leave, health, family, money). voice-dna.md fans out to every instance, so these are FOUNDER-GATED: show the block and ask for one short yes/no covering it, in the same turn, never as a wait-for-approval turn of its own. On a yes, merge via the same PR flow. On a no, drop them.
- After he answers either way, write the current ISO-8601 UTC instant to automation/.voice-delta-answered. That is what stops the 7-day unanswered-delta alert; nothing clears it on its own.

Read the full file on GitHub · 70 lines

Changes

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.

7d ago Changed · +25 lines 9b044f585b43
2mo ago First seen · 45 lines · 59 tokens per session scan A 329ec8961fa0

Subscribe to this mod's changes

voice-refresh is a command published in the GitHub repository assafkip/kipi-system (131 stars, last pushed 2d ago), licensed MIT. It adds 59 tokens to every session and 960 once invoked, about $0.0002 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.

Claude Code Writing and docs agent-framework ai-agents ai-memory knowledge-management markdown

voice-refresh next to 6 commands from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
voice-refreshassafkip/kipi-system	Command	131	2d ago	MIT	A	59	Claude Code
recallmodeled-information-format/mnemonic	Command	23	11d ago	MIT	A	9	Claude Code
querymodeled-information-format/mnemonic	Command	23	11d ago	MIT	A	9	Claude Code
searchmodeled-information-format/mnemonic	Command	23	11d ago	MIT	A	8	Claude Code
gcmodeled-information-format/mnemonic	Command	23	11d ago	MIT	B	7	Claude Code
setupmodeled-information-format/mnemonic	Command	23	11d ago	MIT	B	8	Claude Code
statusmodeled-information-format/mnemonic	Command	23	11d ago	MIT	B	8	Claude Code