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 assafkip/kipi-system/plugin install kipi-coreWrote 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/commands/assafkip/kipi-system/say)<a href="https://agentmods.dev/commands/assafkip/kipi-system/say"><img src="https://agentmods.dev/badge/commands/assafkip/kipi-system/say.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 | $0.00016 | $0.00488 |
| Opus 5 | $0.00008 | $0.00244 |
| Sonnet 5 | $0.00003 | $0.00098 |
| Haiku 4.5 | $0.00002 | $0.00049 |
Grade A, and why
say scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Turn the previous response into audio you can play with speed/seek/pause. No copy-paste into a TTS app.
/say synthesizes the last assistant prose response with OpenAI TTS, writes it
to a stable mp3, and AUTOPLAYS it in a new Terminal window (local, mpv present)
so your keyboard drives speed, seek, and pause — a real terminal takes key
presses, a background player cannot. Over SSH, or without mpv, or with
--no-play, it skips the window and just prints the play command. /say stop
clears any stray playback.
Call the script FIRST, with no preamble text. (The script reads the last prose assistant message from the transcript; any text you emit before the tool call would become that message and get read instead of the founder's intended response.)
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/say-last-response.py" $ARGUMENTS
After it runs, report its stdout verbatim (it prints the autoplay status + the ready-to-run play command). If it exits non-zero, surface stderr verbatim (e.g. the no-key setup line) and do not retry.
Playing it manually (SSH, --no-play, or a second listen)
Autoplay opens a Terminal window for you locally. When you want to drive it yourself instead — over SSH, or to replay — use these:
- Local / in the SSH session (mini speakers):
mpv ~/.config/kipi/say-last.mp3 - Laptop audio over the SSH you already have:
ssh <mini> 'cat ~/.config/kipi/say-last.mp3' | mpv - - Convenience helper:
say-play(local) orsay-play --remote <mini>(laptop)
mpv keys: [ ] speed | <-/-> seek 5s | up/down 60s | space pause | q quit.
Needs brew install mpv (ffplay is a no-speed fallback if mpv is absent).
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 First seen · 41 lines · 16 tokens per session scan A 6113537c677d
say is a command published in the GitHub repository assafkip/kipi-system (109 stars, last pushed today), licensed MIT. It adds 16 tokens to every session and 488 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 2026-08-30.