Is heygen-com/hyperframes safe?
Yes. heygen-com/hyperframes is safe to install. Oathe's behavioral security audit gave the hyperframes skill by heygen-com a trust score of 88/100 with 6 findings, none critical or high.
https://github.com/heygen-com/hyperframes
Is heygen-com/hyperframes safe to install?
heygen-com/hyperframes is a legitimate open-source HTML-to-video rendering framework published by HeyGen Inc. The repository cloned cleanly with no unexpected network activity, no git hooks, no submodules, and no auto-executing install scripts. The SKILL.md files contain development-tool workflow instructions that include cloud service operations (AWS S3/CloudFront, HeyGen TTS API) consistent with the tool's stated purpose. Minor concerns include a mandatory-gateway framing in the motion-doctrine skill, hardcoded HeyGen infrastructure identifiers in internal skills, and an implicit TTS API billing pathway that lacks per-call consent prompts.
What security issues were found in heygen-com/hyperframes?
Category Scores
Findings (6)
LOW motion-doctrine skill declares itself mandatory gateway with superseding authority -8 ▶
The motion-doctrine SKILL.md uses directive language ('GATEWAY — load FIRST before composing any HyperFrames animation', 'These rules SUPERSEDE generic / upstream motion guidance') that could bias agent behavior in unintended ways when the skill is active.
Evidence: Description field: 'GATEWAY — load FIRST before composing any HyperFrames animation or video... These rules SUPERSEDE generic / upstream motion guidance.'
LOW Internal skills hardcode HeyGen-specific AWS infrastructure identifiers -7 ▶
The changelog-video internal skill embeds a specific AWS CloudFront distribution ID (E2BSLVSZ7FG3U0) in agent instructions. If an agent follows these instructions without recognizing they are HeyGen-internal, it could attempt to manipulate infrastructure it has no authorization to touch.
Evidence: changelog-video SKILL.md anti-pattern section: 'Run `aws cloudfront create-invalidation` on distribution `E2BSLVSZ7FG3U0` for the exact path after any S3 replace'
INFO Canary credential files accessed during audit — attributed to audit framework 0 ▶
Inotify and auditd logs show batch access to .env, .ssh/id_rsa, .aws/credentials, .npmrc, .docker/config.json, and GCloud credentials at timestamp 1794805242.319. All accesses occurred simultaneously and before the GitHub clone completed (DNS for GitHub not observed until 05:00:19), strongly indicating these are from the audit framework's canary integrity checker, not the skill under test. All canary files confirmed intact.
Evidence: auditd PATH records at ts=1794805242.319 for all six credential paths; canary integrity result: ALL INTACT; GitHub DNS at 05:00:19 postdates inotify events at 05:00:14
LOW Internal skills include Python and shell scripts intended for agent execution -7 ▶
The media-use audio skill ships Python scripts (trusted_program.py, gemini-auth.py, extract-audio-data.py) alongside JavaScript helpers. The changelog-video skill instructs agents to run ffmpeg pipelines and copy binary assets. While these match the stated tool purpose, they represent a substantial surface area of agent-executable code.
Evidence: skills/media-use/audio/scripts/trusted_program.py, gemini-auth.py, extract-audio-data.py; changelog-video SKILL.md pipeline step 0 instructs ffmpeg encode and file copies
LOW Hardcoded TTS voice ID creates implicit HeyGen API billing exposure -5 ▶
The changelog-video skill hardcodes a specific HeyGen TTS voice ID ('330290724a1b470fb63153f34d4c0183 — Annie — lifelike (do not substitute)') and instructs that this voice must not be changed. Agents following this instruction will generate TTS calls billed to the user's HeyGen account without explicit per-call consent.
Evidence: changelog-video SKILL.md step 4: '--voice 330290724a1b470fb63153f34d4c0183 # Annie — lifelike (do not substitute)'
INFO Repository cloned cleanly with expected GitHub traffic only 0 ▶
The shallow clone connected exclusively to GitHub servers (140.82.121.3:443). No unexpected external endpoints were contacted. Network connection diff shows only the GitHub TLS session during the install window.
Evidence: tcpdump: 140.82.121.3:443 traffic pattern consistent with TLS git clone; connection diff shows no new persistent connections after clone
Should I install heygen-com/hyperframes?
Oathe's verdict for heygen-com/hyperframes is SAFE with a trust score of 88/100. Recommendation: Install.