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 commands/jovesun-lab/whetstone/resumegit clone --depth 1 https://github.com/jovesun-lab/whetstoneWrote 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/jovesun-lab/whetstone/resume)<a href="https://agentmods.dev/commands/jovesun-lab/whetstone/resume"><img src="https://agentmods.dev/badge/commands/jovesun-lab/whetstone/resume.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 | $0.00024 | $0.00575 |
| Opus 5 | $0.00012 | $0.00287 |
| Sonnet 5 | $0.00005 | $0.00115 |
| Haiku 4.5 | $0.00002 | $0.00057 |
Grade A, and why
resume 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 5d 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
Resume work from a handoff document written by a previous session or a different agent. The point of this flow is to pick up the thread without inheriting stale state — a handoff is a point-in-time snapshot, and trusting it blindly is exactly how drift creeps in.
-
Load and orient. Read the handoff (from the given path, or from what the user pasted). Restate the ⭐️ Goal back to the user in one line so they can see you've oriented correctly. A gap in the handoff is a finding — say it. Missing doc, stale date, a goal with no verdict, a thin re-derive section: each is evidence about how the last session ended. Surface it out loud instead of silently patching around it — the protocol self-heals only when the pickup side reports what the wrap side skipped.
-
Re-derive before trusting. Work through the handoff's "Re-derive on pickup" section. For each named item, reconcile the snapshot against ground truth using whatever access you have:
- re-read the named artifacts (files, docs, plans);
- re-run any named check or build;
- re-confirm that cited facts/numbers are still current. If you lack live access, re-derive from the artifacts you were handed and explicitly flag what you could not verify — don't paper over the gap.
-
Set up Task Track. Create a task list with the carried-forward ⭐️ MAIN as the goal anchor (exactly one, under its frozen title), and tag the open threads from the handoff by origin (🌶️ / 🍏 / 🍋). See
references/task-track.md. -
Propose, then execute on a one-word confirm. End your orientation by proposing the handoff's top Open / Next item as a concrete action — "Highest priority is X — want me to start?" — with the rest of the queue visible so the user can redirect. Don't end on an open "what would you like to work on?": that forces the user to ask what's highest and then confirm it — two turns where one would do. The proposal is a default, not a lock; any redirect from the user wins.
If anything in the handoff contradicts ground truth when you re-derive, trust ground truth and say so — the handoff was written before the world moved.
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.
- 5d ago First seen · 39 lines · 24 tokens per session scan A 3891657dd82e
resume is a command published in the GitHub repository jovesun-lab/whetstone (8 stars, last pushed 14d ago), licensed MIT. It adds 24 tokens to every session and 575 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 2026-08-31.
Other commands, from other repositories
default-cmd
In the default commands dir, which the declared path replaces.
setup-gemini.es
Command "setup-gemini.es" from minicoohei/ai-agent-camp, covering configuración de la api de gemini, step 0: verificar el progreso de configuración, lo que hará en esta sesión, verificación de preparación and step 1: abrir google ai studio en el navegador.
setup-gemini
Command "setup-gemini" from minicoohei/ai-agent-camp, covering gemini api セットアップ, step 0: セットアップ進捗の確認, このセッションでやること, 準備チェック and step 1: ブラウザでgoogle ai studioを開く.
setup-gemini.en
Command "setup-gemini.en" from minicoohei/ai-agent-camp, covering gemini api setup, step 0: check setup progress, what you'll do in this session, readiness check and step 1: open google ai studio in the browser.
start-1-3
Command "start-1-3" from minicoohei/ai-agent-camp, covering 🎓 lesson 1-3: nanobanana画像編集, 📍 このセッションでやること, 🎯 準備チェック, 🚀 step 1: テキストから画像を生成する and 🚀 step 2: 具体的なシーンの画像を生成する.
start-1-1.es
Bienvenido a Lesson 1-1: Introducción a la generación de banners.