Skill Trust Decision
minimax-web-search
Hardcoded API key embedded in source code with inadequate security warnings creates critical credential exposure risk.
Most direct threat evidence
Critical Credential Theft
Hardcoded API Key in Source Code A complete MiniMax API key is embedded in plaintext in scripts/mmsearch.py and scripts/mmvision.py. This key grants access to the MiniMax Coding Plan API and can be extracted by anyone with read access to the source code.
scripts/mmsearch.py:11 Why this conclusion was reached
1/4 dimensions flagged Pass
Declared vs actual capability
Declared resources and inferred behavior are broadly aligned.
Review
Hidden execution and egress
1 lower-risk artifacts were extracted and still need context.
Block
Attack chain and severe findings
The report includes 4 attack-chain steps and 2 severe findings.
Review
Dependencies and supply chain hygiene
2 dependency or supply-chain issues need attention.
Attack Chain
01
API key hardcoded in mmsearch.py line 11
exposure · scripts/mmsearch.py:11
02
API key duplicated in mmvision.py line 11
exposure · scripts/mmvision.py:11
03
Attacker extracts key from source code, uses MiniMax API at victim's expense
Impact · scripts/mmsearch.py:11
04
Victim shares skill with exposed credentials in documentation
Impact · SKILL.md:31
What drove the risk score up
Hardcoded API key in source code +35
API key 'sk-cp-_opubDWoTJY5qpJAoI...' embedded in both mmsearch.py and mmvision.py (lines 11 in both)
Credential not protected +15
No attempt to mask or secure the key; exposed in plaintext
Misleading documentation +10
SKILL.md says key is 'built-in' (已内置) presenting it as a feature rather than a security risk
Key exposure surface +8
Key visible in source, commits, and logs if skill is shared/distributed
Most important evidence
Critical Credential Theft
Hardcoded API Key in Source Code
A complete MiniMax API key is embedded in plaintext in scripts/mmsearch.py and scripts/mmvision.py. This key grants access to the MiniMax Coding Plan API and can be extracted by anyone with read access to the source code.
scripts/mmsearch.py:11 Remove the hardcoded key. Read from environment variable MINIMAX_API_KEY with fallback prompt or error. Example: os.environ.get('MINIMAX_API_KEY')
Critical Credential Theft
Duplicate Hardcoded API Key
The same API key is duplicated in scripts/mmvision.py, line 11
scripts/mmvision.py:11 Remove duplicate. Use a shared config or environment variable.
Medium Doc Mismatch
Misleading Security Documentation
SKILL.md presents the hardcoded key as a feature ('已内置在脚本中' - built into scripts) rather than disclosing it as a security risk. Users may unknowingly share this skill with exposed credentials.
SKILL.md:31 Update documentation to warn users: (1) Key must be rotated if exposed, (2) Key should ideally be provided by users, (3) Do not commit this file to public repos
Low Supply Chain
Unpinned Dependency Version
The uvx command runs 'minimax-coding-plan-mcp' without version pinning, which could allow a malicious package update
scripts/mmsearch.py:23 Pin to specific version: uvx minimax-coding-plan-mcp==X.Y.Z
Declared capability vs actual capability
Filesystem Pass
Declared READ
→ Inferred READ
SKILL.md declares file access for scripts Network Pass
Declared READ
→ Inferred READ
Makes API calls to api.minimaxi.com Shell Pass
Declared WRITE
→ Inferred WRITE
subprocess.Popen for uvx command execution Suspicious artifacts and egress
Medium External URL
https://api.minimaxi.com scripts/mmsearch.py:11
Dependencies and supply chain
| Package | Version | Source | Known vuln | Notes |
|---|---|---|---|---|
| uvx | * | pip | No | CLI tool runner, no known CVEs |
| minimax-coding-plan-mcp | * | uvx | No | Version not pinned - potential supply chain risk |
File composition
3 files · 267 lines
Python 2 files · 223 linesMarkdown 1 files · 44 lines
Files of concern · 3
scripts/mmvision.py Duplicate Hardcoded API Key
scripts/mmsearch.py Hardcoded API Key in Source Code · Unpinned Dependency Version · https://api.minimaxi.com
SKILL.md Misleading Security Documentation
Security positives
No reverse shell or C2 infrastructure detected
Network calls limited to legitimate API endpoint (api.minimaxi.com)
No credential exfiltration beyond the self-contained API key
subprocess usage is documented and expected for CLI tool wrappers
No base64 obfuscation or anti-analysis techniques
No sensitive file access (no ~/.ssh, ~/.aws, .env reading)