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/jhamidun/screencast-desktop/recordgit clone --depth 1 https://github.com/JHamidun/screencast-desktopWhat 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 | $0.00035 | $0.00516 |
| Opus 5 | $0.00017 | $0.00258 |
| Sonnet 5 | $0.00007 | $0.00103 |
| Haiku 4.5 | $0.00003 | $0.00052 |
Grade A, and why
record 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 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.
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
Record a demo video of the app: $ARGUMENTS
Work step by step and look at the screen after each one — never act blind.
1. Prepare the stage. desktop_windows — find the window. Put it on a
non-primary monitor if there is one (desktop_monitors) so it stays off the
working screen: desktop_place_window(..., fit=0.7). Confirm from the
response that the window fits entirely — an app may come back with a size
other than the one it was given, and then it will be cropped in the recording.
2. Check the frame for private content. Take a desktop_screenshot and
look: mail, private conversations, names, passwords, notifications. If there
is anything in the frame that should not be in the video — tell me about it
before recording, not after.
3. Scout the route LIVE. Walk the scenario with clicks via ui_snapshot /
desktop_click, confirming from the returned shots that you hit the right
places and that the app responds as expected. Mistakes are fine here — this is
a draft, it is not being recorded.
4. Return the app to its initial state. An unclosed search box or a card left open from scouting will end up in the video and ruin it.
5. Write the voiceover for the route you found. The script is about actions, not content: a live list changes between the draft and the recording. Show me the script before recording.
6. Record the clean take. desktop_record_start → actions →
desktop_record_stop. Between actions, hold pauses matching the length of the
narration lines.
7. Assemble and watch. desktop_render, then open the finished file and
check with your own eyes: did the camera go where it should, are there black
bars at the edges, is the cursor visible. A coordinate mistake produces a
technically valid file in which the camera looks at the wrong place.
If something is off — rebuild with desktop_render using different
parameters: the app is not relaunched and the screen does not need to be
re-recorded.
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 · 45 lines · 0 tokens per session scan A d6cd8124545d
record is a command published in the GitHub repository JHamidun/screencast-desktop (0 stars, last pushed 7d ago), licensed MIT. It adds 35 tokens to every session and 516 once invoked, about $0.0002 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
implement
yes, thank you! please proceed with implementation.
suggest-solution
if there is better solution, please suggest it.
create-forwarding-rule
Create a ZPA client forwarding policy rule to bypass or intercept traffic.
feature
End-to-end feature/bug-sweep workflow for ui-debugger-mcp — understand, reproduce against a real target, explore in parallel, split into path-disjoint slices, build with a hive of agents in this ONE checkout (never worktrees), gate green, PR, merge, release to npm. Tracks in GitHub issues. Reads intent from the prompt.
auto
AI auto strategy executor. Analyzes task and automatically selects and executes the best strategy: direct execution, hybrid-auto PRD generation, hybrid-worktree isolated development, or mega-plan multi-feature orchestration.
spec-plan
Run a resumable spec interview to produce spec.json/spec.md (planning-time), then optionally compile to prd.json. Usage: /plan-cascade:spec-plan " " [--output-dir ] [--flow ] [--first-principles] [--max-questions N] [--feature-slug ] [--compile] [--tdd ] [--confirm] [--no-confirm].