sidenote

A command for recording a passing thought as a later task without acting on it now. The thought is saved in a project file so it remains available after the current work ends.

In plain words
What is it for?
Use it to log follow-up ideas, investigations, or fixes that should be handled separately.
Why use it?
It prevents useful ideas from interrupting the task in progress while preserving them for future work.

Command

Part of the greybeard plugin — 2 skills, 2 commands, 2 hooks shipped together

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.

agentmods
npx agentmods add commands/clawnify/greybeard/sidenote
Clone the repo
git clone --depth 1 https://github.com/clawnify/greybeard

Or install greybeard, the plugin that ships this one along with the rest of its 2 skills, 2 commands, 2 hooks.

Per session 21 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 735 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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.

ModelPer sessionOnce invoked
Fable 5 $0.00021 $0.00735
Opus 5 $0.00010 $0.00367
Sonnet 5 $0.00004 $0.00147
Haiku 4.5 $0.00002 $0.00073

Measured 2d ago against content hash 3cff2cc6b996, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

sidenote 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 2d 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.

commands/sidenote.md · 30 lines

What it actually says

Park this thought — $ARGUMENTS — as a sidenote. A sidenote is a later-task, not a now-task. Your job is to record it and keep going, nothing more.

The contract — hold it exactly:

  • Do not act on it. Don't investigate it, don't open the files it mentions, don't add it to your current plan, don't answer the question it poses. It is parked, not pending.
  • Do not let it touch the current task. Its scope, plan, priorities, and pace stay exactly as they were. In particular: do not speed up or cut corners to "get to" the note. The whole point of parking it is that it costs the current work nothing.
  • Urgency doesn't override the contract. If the note reads as urgent ("prod is down", "this is broken"), that's the wrong channel — a real emergency is a normal interruption, not a sidenote. Still just log it and resume; the human will escalate directly if they meant to.

Persist it so it survives this session (context gets compacted; memory doesn't count as "logged"):

  1. Append one entry to .claude/sidenotes.md at the project root — create the file (and .claude/) if missing.

  2. Format — an open checkbox, the verbatim thought, then a one-line anchor so the note still makes sense when another session or agent reads it cold (the verbatim thought alone is full of "this / here / it" that dangles without the moment that produced it):

    - [ ] [<today's date, YYYY-MM-DD>] <the thought, captured verbatim>
      ↳ <task you were mid-way through> · <file/area the note points at, if "this"/"here" resolves to one> · <branch>@<short-sha>
    
    • Thought — verbatim. Don't paraphrase or "improve" it; capture what was said.
    • Anchor — coordinates you already hold, not a research task. Fill it only from what's already in front of you: the task in progress, the file/area currently in focus, the current branch and short SHA (already in your session's git context). Do not open files, grep, or investigate the note's subject to complete it — that would break the park contract. Any field you don't already know, omit. The anchor records where you were standing; it never sends you looking.

Then: one-line acknowledgement (e.g. Parked in .claude/sidenotes.md — continuing.) and resume exactly where you left off, as if the note had never arrived.

If $ARGUMENTS is empty (a bare /sidenote), there's nothing to park — treat it as a flush: read .claude/sidenotes.md and list the open entries back to me so I can decide what to pick up. Open means not checked off — every entry except the ones marked done (- [x]), so older bullets without a checkbox still count. Change nothing, act on nothing, then resume. (Checking an item off to - [x] happens when it's actually handled — not part of parking or flushing.)

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.

  1. 2d ago First seen · 30 lines · 21 tokens per session scan A 3cff2cc6b996

Subscribe to this mod's changes

sidenote is a command published in the GitHub repository clawnify/greybeard (6 stars, last pushed 10d ago), licensed MIT. It adds 21 tokens to every session and 735 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.