📦

Audit History

douyin-video-copy-extract - 2 audits

Version comparison

Capability and finding changes across audited versions, newest first.

VersionDateResultReview itemsChange vs previous
v2 LatestNov 6, 2026, 04:58 AM 1 confirmed1No capability change
v1 Nov 2, 2026, 12:22 PM 1 confirmed1Baseline

Nov 6, 2026, 04:58 AM

Most static alerts match Markdown, readable Chinese prose, or a credential check that reports only presence. The CLI examples execute an unpinned npm package with access to the agent environment. Polling lacks a time or attempt limit; no evidence found of credential exfiltration, encoded payloads, or audit-directed prompt injection.

2
Files scanned
142
Lines analyzed
5
Review items
0
False positives ignored

Confirmed security concerns (1)

Medium
Polling Has No Execution Budget
The rules require polling until terminal status, tool failure, or user cancellation. A perpetually nonterminal job can consume ongoing requests and agent execution.
The text explicitly lists only three stopping conditions and gives no attempt or elapsed-time limit. Poll delays reduce request frequency but do not bound execution.
Capability review items (1)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

Medium
Ruby/shell backtick execution
```bash
The fenced examples run npx -y socialdatax-skills@latest, allowing changing registry code to execute without an installation prompt. That code inherits the agent environment, including service credentials.
Audited by: codex

Nov 2, 2026, 12:22 PM

Of 54 static alerts, 53 are false positives involving Markdown, ordinary Chinese text, service links, or documented authentication. One alert covers real execution of an unpinned npm package; a separate semantic finding identifies polling without a deadline or attempt limit. No evidence found of prompt injection, credential exfiltration, or reconnaissance in the supplied files; external package and service behavior was not verified.

2
Files scanned
126
Lines analyzed
5
Review items
0
False positives ignored

Confirmed security concerns (1)

Medium
Polling Has No Time or Attempt Limit
The workflow continues querying every nonterminal job and permits stopping only at completion, tool failure, or user request. A stalled service can consume unbounded requests and agent execution time.
The stopping conditions explicitly omit a deadline or attempt budget. Waiting between requests reduces frequency but does not bound total resource consumption.
Capability review items (1)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

Medium
Ruby/shell backtick execution
```bash
The fenced examples execute npx -y socialdatax-skills@latest, automatically running mutable third-party code with access to the configured API credential. This is a supply-chain execution risk, not Ruby backtick substitution.
Audited by: codex