A procedure for recording and reviewing short desktop videos that demonstrate a user interface or test-harness interaction. It covers cases where a still image cannot show motion, timing, sound, or a sequence of clicks.

Use it to prove actions such as clicking controls, opening inspectors, checking a standalone or plug-in host, or comparing an interface with reference material.

It provides evidence that an interaction works and helps reviewers compare the implementation with its intended source. It also keeps video tools and generated files separate from the main application.

video-proof — Skill for Claude Code by Generous-Corp: A procedure for recording and reviewing short desktop videos…

Install

Getting it into your agent

It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.

Clone the repo

git clone --depth 1 https://github.com/Generous-Corp/pulp

agentmods

npx agentmods add skills/generous-corp/pulp/video-proof

Made for: Claude Code.

Or install pulp, the plugin that ships this one along with the rest of its 63 skills, 30 commands, 3 hooks, 1 MCP server.

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/skills/generous-corp/pulp/video-proof.svg)](https://agentmods.dev/skills/generous-corp/pulp/video-proof)

Your own site

<a href="https://agentmods.dev/skills/generous-corp/pulp/video-proof"><img src="https://agentmods.dev/badge/skills/generous-corp/pulp/video-proof.svg" alt="Measured on agentmods" height="20"></a>

Per session 97 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 11,157 The whole file, excluding the scripts and references it only reads on demand.

Security scan C 1 finding. 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.

Model	Per session	Once invoked
Fable 5.1	$0.00097	$0.11157
Opus 5	$0.00048	$0.05579
Sonnet 5	$0.00019	$0.02231
Haiku 4.5	$0.00010	$0.01116

Measured yesterday against content hash 54bc1319fce3, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade C, and why

video-proof scanned grade C with 1 finding 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 yesterday.

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.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

rm -rf "$tmp_home"

.agents/skills/video-proof/SKILL.md · 1,046 lines

How it starts

The opening of the file, as written. The whole thing — 1,046 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Validation video proofs

Use this skill when a screenshot is not enough to prove a UX interaction: clicking a control, opening an inspector, checking a standalone/plugin host flow, or showing an implementation next to source material. Default to still captures for static assertions; record video only when motion, timing, audio/visual state, or reviewer comprehension matters.

Rules of engagement

Work on a feature branch or dedicated worktree until humans have reviewed the proof workflow. Do not merge generated videos or local config.
Treat video proof capture as optional developer tooling, not a core runtime dependency and not a normal pulp add project package. Prefer the pulp tool install video-proof add-on for user-facing setup; it owns the ffmpeg/Remotion bootstrap and large or license-sensitive dependencies. Keep direct source-tree setup (npm --prefix tools/local-ci install) as the feature-branch iteration fallback.
Use pulp tool info video-proof --json when an agent or setup script needs the machine-readable install policy. It should report a machine-scoped tool_addon, package_format: not_pulp_add, and, on this feature branch, artifact_status: source_tree_iteration. It should also report artifact_pack_command, artifact_pack_npm_script, artifact_verify_command, and artifact_manifest_schema so the artifact-review step is discoverable from the registry.
Use python3 tools/local-ci/pack_video_proof_tool.py --json when the branch needs a reviewable source add-on artifact. The packer records sha256, size, included files, npm pins, and policy flags, and it excludes node_modules, Remotion caches, and generated videos. Follow it with python3 tools/local-ci/pack_video_proof_tool.py --verify <manifest> --json before treating the artifact as reviewable.
Prefer short clips: 5-12 seconds is the normal range. Keep the recording focused on the window or component under test.
Prefer video/proof-composed.mp4 for review. Keep video/proof.mp4 as the raw debugging artifact.
Treat size as a test result. Use --video-attachment-budget-mb 100 for paid/pro GitHub issue attachment planning only when the uploader is eligible for videos above 10 MB. Otherwise use a 10 MB budget or publish a local report and link the served report instead of attaching the MP4.
For review issues that may be handled by a non-pro uploader, pass --small-video --small-video-budget-mb 10 to desktop video, or rerender with desktop compose-video --small-video --small-video-budget-mb 10, so video/proof.small.mp4 is available as a convenience attachment.
Never commit tools/local-ci/config.json, tools/local-ci/node_modules/, .remotion/, or generated desktop artifact bundles.

Read the full file on GitHub · 1,046 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.

yesterday First seen · 1,046 lines · 97 tokens per session scan C 54bc1319fce3

Subscribe to this mod's changes

video-proof is a skill published in the GitHub repository Generous-Corp/pulp (16 stars, last pushed yesterday), licensed MIT. It adds 97 tokens to every session and 11,157 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-04.

Design and UX E2E testing Testing

qa-test-planner

Generate comprehensive test plans, manual test cases, regression test suites, and bug reports for QA engineers. Includes Figma MCP integration for design validation.

davila7/claude-code-templates · 34 tokens

visual-validation

Every UI change should be visually verified before it ships. Peekaboo captures pixel-accurate screenshots. The system compares before/after and flags visual regressions. No manual "looks good to me" — the machine verifies what the machine built.

alinaqi/maggy · 0 tokens

design-ship

One-shot pipeline turning a claude.ai/design link into a pull request: scaffold via /ork:design-import, stories and specs via /ork:cover, browser verification via /ork:expect, then open the PR. Use when a design link should come back as a PR with no intermediate steps; if all you need is the components written to…

yonatangross/orchestkit · 84 tokens

shaft-accessibility-testing

Use when testing web or mobile accessibility for WCAG conformance, keyboard and focus behavior, semantics, errors, or user impact.

ShaftHQ/SHAFT_ENGINE · 31 tokens

visual-playtest

Visually inspect a local or deployed site/app in a real browser and report concrete layout, responsive, or design regressions. Use when asked to visually check UI changes, reproduce a screenshot issue, compare a page with a reference, or playtest representative visual states. Do not use for code-only review or…

swyxio/skills · 71 tokens

argos-cli

Operate Argos visual testing from the terminal with the argos CLI — inspect builds and snapshot diffs, submit reviews, request reviewers, post comments, inspect a test's flakiness and its recurring changes, ignore flaky test changes, configure a project and its contributors, manage a team's members, invites and email…

argos-ci/argos-javascript · 143 tokens

video-proof