Security audit
Security checks for vulnerabilities and agentic risk
ClawScan marks this game trailer creation skill suspicious because it reports risky file handling and a path validation flaw, while inspection of some files was incomplete.
Consider avoiding installation until the directory deletion and path exposure issues are reviewed and corrected. If you proceed, use a disposable environment with no important files, and remember that six files were only partially inspected.
scripts/capture.tsStatic-server path validation permits access to sibling directories
Disclosure of readable files in matching-prefix sibling directories. Page JavaScript can read the response from its own origin, exposing data outside the selected game directory.
Referenced artifact was not completely inspected
For your own game: copy `examples/lantern-canyon/director.json` next to the game's `index.html`, point
Referenced artifact was not completely inspected
{ "url": "index.html", "fps": 30, "width": 1920, "height": 1080, "seed": 7, "bpm": 100,
Referenced artifact was not completely inspected
| `scripts/director-runtime.js` | in-page half: trailer camera override, overlays, hidden selectors, letterbox, events |
Referenced artifact was not completely inspected
| `scripts/cdp.ts` | minimal Chrome DevTools Protocol client (launch, inject, keys, mouse, screenshot) |
Referenced artifact was not completely inspected
| `scripts/capture.ts` | director JSON → frames + manifest; `--stills`, `--only`, `--verify`, `--gl swiftshader` |
Referenced artifact was not completely inspected
| `scripts/cut.ts` | frames → master MP4: order, fades, grade, mix, two-pass loudnorm |
Referenced artifact was not completely inspected
| `scripts/check.ts` | duplicates, freezes, black frames, loudness, contact sheet, text/IP/claim audit |
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<meta charset="utf-8">
<title>Trailer card</title>
<link rel="icon" href="data:,">
<!-- A title / end card for code-game-trailer. Use it as a director shot: { "card": "cards/card.html?title=...", "beats": 3 }.
Query: kicker, title, sub, foot (all optional text), accent (hex without #, default ff5315), bg (hex, default 030507),
embers=0 to switch the drifting embers off. The CSS animations and the canvas run on the recorder's virtual
clock, so every frame is exact. Change the font stack for your game; system fonts need no loading. -->
Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
| text | only your game's words | debug words, franchise names, claims without proof are flagged |
The example's final run: 0 duplicates and 0 black frames in all gameplay shots, -13.7 LUFS, -1.4 dBFS
true peak, 1920x1080 30 fps H.264 + AAC, no warnings. Then look at the sheet: the first sheet of this
example passed every number and still showed a three-second opening in which the hero never entered
the frame.
Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
if (CLAIM.test(text)) warn(`${where}: claim "${text.match(CLAIM)![0]}": keep it only if NOTES.md proves it`);
}
await Bun.write(video.replace(/\.mp4$/, '') + '.check.json', JSON.stringify(report, null, 1));
console.log(report.warnings.length ? `${report.warnings.length} warning(s); now open the sheet and look at it` : 'no warnings; now open the sheet and look at it');
The skill uses 'network' capability that is not covered by the declared 'allowed-tools' preapproval. Review the host runtime's actual enforcement policy; runtime permissions are unknown unless separately verified, and this declaration alone does not establish a permission ceiling.
The skill uses 'env' capability that is not covered by the declared 'allowed-tools' preapproval. Review the host runtime's actual enforcement policy; runtime permissions are unknown unless separately verified, and this declaration alone does not establish a permission ceiling.
npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.
Recording recursively deletes the computed shot directory without confirmation or verifying that it contains only generated frames. Because both the output location and shot ID come from the director configuration, a malicious or mistaken configuration can use traversal components such as '..' to delete directories outside the intended capture area with the running user's privileges.