Back to skill

Security audit

remotion-to-hyperframes

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious with medium confidence, mainly because it uses unpinned commands and dependencies that may download and run changing third-party code; no malicious intent is established.

Consider avoiding installation until you can verify the package versions and review the skill's external requests and execution steps. If you proceed, use a restricted environment without sensitive files or credentials. The scan was partial: 70 of 72 components were scanned, some references could not be resolved, and one static analyzer was degraded.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (48)

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, file_write, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

This is a CLI invocation, not an MCP server, but the unversioned execution risk is real: if no suitable local installation exists, npx can download and execute a registry-resolved version. A compromised release could execute with the rendering user's privileges; the referenced plugin rules might mitigate this, but their contents are not supplied.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

Although catalog lookup is a legitimate task, unversioned npx can obtain and execute a package before performing the search. The security concern is supply-chain code execution, not MCP configuration or the catalog query itself; applicable plugin installation safeguards are not available for review.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The documented test command uses an unversioned CLI invocation that can download and execute a registry package when a local executable is unavailable. Its role as a rendering fixture reduces expected exposure but does not isolate a compromised package from the developer's credentials or files.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The rendering instructions invoke an unversioned npm CLI, not an MCP server. If the executable is unavailable locally, npx may download and execute a registry-selected release with the user's privileges, creating supply-chain exposure even in a test fixture.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

This unversioned npx command can resolve and execute an unreviewed HyperFrames release when no local installation supplies the CLI. Rendering is legitimate, but a compromised dependency would inherit the invoking user's access rather than being confined to producing video.

Content

No source excerpt is available for this finding.

Output overwriting is enabled without disclosure that existing files may be replaced.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

Config.setOverwriteOutput(true) allows subsequent renders to replace existing output files without confirmation. The surrounding comments explain rendering fidelity settings but do not disclose the risk of replacing existing user data.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · assets/test-corpus/tier-4-escape-hatch/cases/03-async-metadata.tsx (reported line 32)May include surrounding context.

tsx
export const calculateMetadata = async ({ props }: { props: Props }) => {
  const response = await fetch(
    `https://api.example.com/duration?text=%24%7BencodeURIComponent%28props.text%29%7D%60,
  );
  const { durationInFrames } = await response.json();
  return {

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · references/api-map.md (reported line 68)May include surrounding context.

md
See [media.md](media.md) for trim, volume ramps, and decoder notes.

| Remotion                               | HyperFrames                                                                                                                                                                                                                                                                                                                                  |
| -------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `<Audio src volume>`                   | `<audio data-start data-duration data-track-index data-volume src>`

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · references/api-map.md (reported line 70)May include surrounding context.

md
|
| -------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `<Audio src volume>`                   | `<audio data-start data-duration data-track-index data-volume src>`                                                                                                                                                                                                                                                                          |
| `<Audio playbackRate startFrom endAt>` | `data-playback-rate`, `data-trim-start`, `data-trim-end`

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · references/api-map.md (reported line 71)May include surrounding context.

md
-------------------- |
| `<Audio src volume>`                   | `<audio data-start data-duration data-track-index data-volume src>`                                                                                                                                                                                                                                                                          |
| `<Audio playbackRate startFrom endAt>` | `data-playback-rate`, `data-trim-start`, `data-trim-end`                                                                                                                                                                                                                                                                                     |
| `<Video src>`                          | `<video playsinline data-has-audio="true" data-start data-duration data-track-index src>` only when the file has an audio stream (check with `ffprobe -v error -select_streams a -

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · references/api-map.md (reported line 73)May include surrounding context.

md
|
| `<Video src>`                          | `<video playsinline data-has-audio="true" data-start data-duration data-track-index src>` only when the file has an audio stream (check with `ffprobe -v error -select_streams a -show_entries stream=codec_type -of csv=p=0 <file>`: non-empty means it has audio) and the source is not `muted` or `volume={0}`; otherwise `muted` instead |
| `<OffthreadVideo>`                     | `<video>` — HF doesn't need the off-thread variant (uses headless Chrome)                                                                                                                                                                                                                                                                    |
| `<Img src>`                            | `<img>`

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · references/api-map.md (reported line 74)May include surrounding context.

md
the source is not `muted` or `volume={0}`; otherwise `muted` instead |
| `<OffthreadVideo>`                     | `<video>` — HF doesn't need the off-thread variant (uses headless Chrome)                                                                                                                                                                                                                                                                    |
| `<Img src>`                            | `<img>`                                                                                                                                                                                                                                                                                                                                      |
| `<IFrame src>`                         | `<iframe>` — HF auto-falls back to screenshot mode for nested iframes

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · references/api-map.md (reported line 75)May include surrounding context.

md
|
| `<Img src>`                            | `<img>`                                                                                                                                                                                                                                                                                                                                      |
| `<IFrame src>`                         | `<iframe>` — HF auto-falls back to screenshot mode for nested iframes                                                                                                                                                                                                                                                                        |
| `staticFile("x.png")`                  | `"assets/x.png"` in the root, `"../assets/x.png"` inside a `compositions/` scene file — copy the file into `hf-src/assets/` next to `index.html`

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · references/api-map.md (reported line 77)May include surrounding context.

md
|
| `staticFile("x.png")`                  | `"assets/x.png"` in the root, `"../assets/x.png"` inside a `compositions/` scene file — copy the file into `hf-src/assets/` next to `index.html`                                                                                                                                                                                             |
| `delayRender()` / `continueRender()`   | drop — HF waits on asset readiness via the Frame Adapter pattern                                                                                                                                                                                                                                                                             |

## Transitions

MCP server referenced without pinned version: 'npx esbuild'.

Medium
Category
MCP Rug Pull
Confidence
87% confidence
Finding

This is an esbuild CLI invocation, not an MCP server, but the unpinned execution concern is valid: npx can download and execute a package when no suitable local installation exists. A compromised resolved release would run with the agent's filesystem and credential access; no malicious package is demonstrated here.

Content

No source excerpt is available for this finding.

The evaluation workflow installs project dependencies without warning that installation can execute project and dependency scripts.

Medium
Category
Output Handling
Confidence
97% confidence
Finding

npm install can execute project and dependency lifecycle scripts, and npm run render explicitly executes project-defined commands. Although this section describes fixtures, applying the workflow to an untrusted imported project can execute attacker-controlled code with the agent's privileges without an explicit trust boundary.

Content

No source excerpt is available for this finding.

Private credentialed assets are to be downloaded and embedded without warning about persistent copies.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

Downloading and embedding credential-protected images creates persistent copies that no longer inherit the original server's access controls. If the generated project is shared, published, or committed, private source assets can be disclosed; the instructions do not address that change in confidentiality.

Content

No source excerpt is available for this finding.

The generated Lottie example loads executable third-party CDN code without a warning.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

The generated HTML executes CDN-hosted JavaScript without integrity verification. Although the URL pins a version, a compromised CDN response could execute in the composition's browser context and access render content or same-origin resources; this does not by itself establish host-level execution.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The command invokes a CLI rather than an MCP server, but npx may fetch and execute an unpinned HyperFrames release. A compromised resolved package could execute with the agent's privileges, making the missing dependency constraint relevant even though adding a transition is legitimate.

Content

No source excerpt is available for this finding.

The shader-transition workflow instructs the agent to run an npm package command without warning about third-party code execution or project changes.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow directs the agent to execute a potentially downloaded npm CLI without identifying the code-execution trust boundary. Adding a transition necessarily changes the project and is expected, but executing an unreviewed package with ambient agent privileges is the substantive risk; this overlaps the RP1 findings.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

This repeated command has the same unpinned package-execution risk as the table example, despite not referencing an MCP server. When no local CLI is available, npx can resolve remote executable code whose version has not been reviewed.

Content

No source excerpt is available for this finding.

The comparison strip repeats the first sampled frame in every row instead of displaying different timestamps as documented.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The split filters duplicate the entire selected-frame stream rather than assigning a different frame to each row, and the single output frame therefore contains repeated copies of the first selected pair. This is a validation-integrity defect: timestamps imply broader coverage than the image provides, allowing later visual defects to escape manual review, although it does not bypass the separate SSIM comparison.

Content

No source excerpt is available for this finding.

Catalog searches require English queries without user opt-in or a documented justification.

Low
Category
Output Handling
Confidence
98% confidence
Finding

The required catalog command explicitly specifies that the visual description must be in English. This imposes a natural-language requirement without offering a language choice or explaining an English-only catalog limitation.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
77% confidence
Finding

The CLI dependency accepts later 4.x releases, allowing an unconstrained installation to select code that was not reviewed with this fixture. This is a conditional supply-chain weakness rather than evidence of compromise; a committed and enforced lockfile, which is not supplied here, could substantially mitigate it.

Content

Scanner excerpt · assets/test-corpus/tier-1-title-card/remotion-src/package.json (reported line 9)May include surrounding context.

json
"render": "remotion render TitleCard out/baseline.mp4"
  },
  "dependencies": {
    "@remotion/cli": "^4.0.0",
    "react": "^18.3.1",
    "react-dom": "^18.3.1",
    "remotion": "^4.0.0"