Security audit
Security checks for vulnerabilities and agentic risk
The skill appears to be a simple local AI incident ledger, but its metadata understates file and CLI execution needs while the ledger may store sensitive incident details.
Review this before installing in a security or compliance workflow. Use only a deliberate ledger path, avoid recording secrets or raw breach evidence unless that location is approved, and treat the current permission metadata as inaccurate because the skill does need local file access and CLI execution. The publisher should align the manifest, documented flags, and storage warnings before routine use.
The skill claims filesystem: false and subprocess: false, yet its purpose and examples explicitly require creating, reading, and summarizing a JSONL ledger from disk via a Python script. This permission mismatch can mislead users and orchestrators into granting trust under false assumptions, while the implementation may still attempt disk access outside the declared safety model.
This file contains agent-facing instructions in Chinese and explicitly states it is for AI coding agents and maintainers. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation when no alternative language option or opt-in is provided.
The documented interface instructs users to invoke --record, --list, and --report, while the manifest declares an args entry point with a required command of add|list|report. This mismatch can cause operators or agents to invoke the tool incorrectly, bypass expected control paths, or build unsafe wrappers around the wrong interface, reducing reliability of incident logging during security events.
The usage text encourages one-click incident recording but does not prominently warn that records are written to a persistent JSONL file on disk, potentially containing sensitive incident details, model names, actions, or breach evidence. In security and compliance contexts, undisclosed persistence can create privacy, retention, and evidence-handling risks if users assume the action is ephemeral.
The module docstring, CLI description, and help text are presented entirely in Chinese, which imposes a specific language on users. The policy requires flagging language or locale constraints unless the skill offers opt-in or clearly justifies the locale restriction.
SQP-3 applies to all file types and includes language or locale policy violations. The file describes a bilingual rewrite and provides only English and Chinese content, but does not state that users can choose their preferred language or that the locale limitation is required by the skill's context.
The natural-language content of this file is entirely in Chinese, including the core description and usage summary, with no indication that users or agents can choose another language. This can violate a language/locale policy when a skill implicitly requires Chinese comprehension without documenting opt-in or a justified regional constraint.
