doc-search

Security checks across malware telemetry and agentic risk

Overview

The skill is purpose-aligned for local document search, but it auto-loads a local pickle database, which can be unsafe if the stored data is tampered with, and it persists document/query content on disk.

Install/use this only in trusted projects, be aware that indexed document text is stored under ./chroma_data, and remove or clear that directory when you do not want the data retained. Do not run it in directories containing untrusted chroma_data pickle files.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

#
ASI05: Unexpected Code Execution
Medium
What this means

If an attacker or untrusted project places a malicious documents_data.pkl file in the expected ./chroma_data location, using the skill could run unintended code in the user's environment.

Why it was flagged

The vectorizer automatically loads a pickle file from its persistence directory when initialized. Python pickle deserialization can execute code if the file is malicious or has been tampered with.

Skill content
self._load_from_disk()
...
with open(self.data_file, 'rb') as f:
    data = pickle.load(f)
Recommendation

Only use this skill in trusted directories, delete unexpected ./chroma_data/*.pkl files, and prefer a safer storage format such as JSON/SQLite or add integrity checks before loading pickle files.

#
ASI06: Memory and Context Poisoning
Low
What this means

Private document snippets and search state may remain on disk and influence later searches until the collection is cleared or the data directory is removed.

Why it was flagged

The documentation clearly states that vectorized document data and indexes are persisted and automatically reloaded.

Skill content
向量数据:`chroma_data/documents_data.pkl`
文件索引:`chroma_data/documents_index.json`
自动保存:每次添加/删除文件后自动保存
自动加载:初始化时自动恢复数据
Recommendation

Use a dedicated persistence directory, avoid indexing sensitive files unless retention is acceptable, and clear or delete the chroma_data directory when the stored knowledge base is no longer needed.