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 skills add PerpetualSoftware/pad --skill capturegit clone --depth 1 https://github.com/PerpetualSoftware/padWrote 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/perpetualsoftware/pad/capture)<a href="https://agentmods.dev/skills/perpetualsoftware/pad/capture"><img src="https://agentmods.dev/badge/skills/perpetualsoftware/pad/capture/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.
<a href="https://agentmods.dev/skills/perpetualsoftware/pad/capture"><img src="https://agentmods.dev/badge/skills/perpetualsoftware/pad/capture.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00010 | $0.00517 |
| Opus 5.5 | $0.00004 | $0.00207 |
| Sonnet 5.5 | $0.00002 | $0.00103 |
| Haiku 4.5 | $0.00001 | $0.00052 |
Grade A, and why
capture 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 1mo 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
If the pad CLI is missing (command not found), say so and point at
https://getpad.dev/install — don't retry or guess at alternate paths. If no
.pad.toml is found in the directory tree, first run pad auth whoami
(fast, safe, never blocks): if it shows a real user, this machine is already
set up — run pad workspace init to link and create the workspace
non-interactively, then continue capturing. If it reports anything else (not
configured, session expired), this machine hasn't finished setup — don't run
pad init/pad workspace init yourself. Both now fail fast when
stdin/stdout aren't terminals instead of hanging (BUG-2538/BUG-2577/
BUG-2592) — though on a harness whose stdin AND stdout look like a real
terminal they still drop into the browser flow — and either way the outcome
just means a human needs an interactive terminal, which doesn't help you. Tell the user to run pad init themselves in an
interactive terminal (in Claude Code, suggest ! pad init) and stop.
Before creating, run pad item list conventions --field trigger=always --field status=active --format json --full and apply whatever it returns —
always-on conventions are project rules the workspace has made mandatory
(privacy, sizing, format, etc.), and low-ceremony capture means skipping the
full bootstrap weight (dashboard, playbooks, roles), not skipping mandatory
rules. This is the only context load this skill does; it's deliberately
narrower than the main skill's full bootstrap.
Capture $ARGUMENTS into the Pad workspace as the item type it most
resembles (idea, task, bug, note), following any conventions just loaded.
Run pad collection list --format json to see the workspace's collections,
pick the best fit, then show what you're about to create (collection, title,
content) and confirm before running pad item create <collection> "<title>" --content "...". Keep the title short; put nuance in the content.
Exception: if any loaded convention explicitly opts into autonomous capture, skip the confirmation — create immediately and confirm with the new issue ID in one line, no ceremony. Confirm-first is the default for everyone else.
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.
- 1mo ago First seen · 40 lines · 10 tokens per session scan A 19c3e2c67b75
capture is a skill published in the GitHub repository PerpetualSoftware/pad (195 stars, last pushed today), licensed Apache-2.0. It adds 10 tokens to every session and 517 once invoked, about $0.0000 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.
Other skills, from other repositories
daily-brief-loop
Use only when the user explicitly asks to run, sync, advance, or persist Mora's durable daily brief. Executes the once-per-day, human-checkpointed NO-SEND operator loop, performs configured source syncs, writes the dated artifact and run log, and advances the brief watermark.
mora-recall
Retrieve the user's own history before answering questions about their people, projects, meetings, messages, decisions, commitments, or past work. Use for requests such as "what did we discuss", "what did I decide", "what do you know about", "catch me up", and "what changed".
dining-concierge
Use when recommending a restaurant, bar, cafe, happy hour, or other food or drink outing—especially when the request names people the user knows. Grounds live recommendations in task-minimal Mora history before using public web information.
memory-capture
Save durable information to Mora only when the user explicitly asks to remember, save, or capture a fact, decision, insight, or task. Use for governed write-back, not for implicit extraction from ordinary conversation and never for deletion.
haypile
Search the user's local documents (PDF, docx, pptx, markdown, text, HTML, mbox email) through Haypile, a local search daemon. Use when the user asks what their files say, wants passages from contracts, papers, or notes, or wants a folder indexed for search. Every result carries a file and page citation. Runs entirely…
gh-skill
Manage agent skills with gh skill. Use this skill to discover, preview, install, update, and publish Agent Skills so an agent can self-manage the skills available in its environment.