Skill Trust Decision
daily-memory-summary
The skill performs declared clipboard/notification aggregation but contains undeclared contact identity extraction with corporate targeting patterns that creates a significant privacy concern.
Why this conclusion was reached
0/4 dimensions flagged Pass
Declared vs actual capability
Declared resources and inferred behavior are broadly aligned.
Review
Hidden execution and egress
2 lower-risk artifacts were extracted and still need context.
Pass
Attack chain and severe findings
There is no explicit malicious chain in the report.
Review
Dependencies and supply chain hygiene
Dependency information is incomplete, so supply-chain confidence stays limited.
What drove the risk score up
Undeclared contact extraction +20
SKILL.md does not mention contact parsing; skill-card.md only vaguely references 'contact parsing behavior beyond the basic daily summary description'
Corporate targeting patterns +10
Regex patterns explicitly target Lenovo corporate identity (联想/DT/Brad) and corporate communication tools (teams, outlook, wechat)
Hard-coded user path +5
/Users/wutianfu/ path is hard-coded, not configurable per skill-card.md's own warning
Benign declared functionality +-5
Core clipboard/notification aggregation behavior is documented and non-malicious
Most important evidence
Medium Doc Mismatch
Contact parsing not declared in SKILL.md
SKILL.md states the skill 'reads clipboard and notification data, writes daily memory' but does not mention contact identity extraction. The script (lines 35-47) iterates through notification sources (teams, outlook, wechat) and extracts email addresses, names, and organizational identifiers.
scripts/summarize.sh:35-47 Either document this behavior explicitly in SKILL.md or remove the contact parsing block
Medium Sensitive Access
Corporate identity extraction with targeted patterns
The regex patterns explicitly target Lenovo corporate environment: lenovo.com emails, Chinese characters (联想), organizational labels (DT, Brad), and Chinese team terms (团队). This suggests deliberate targeting of corporate communications for identity harvesting.
scripts/summarize.sh:43 Review whether this corporate identity extraction is intentional and authorized
Low Priv Escalation
Hard-coded absolute path not configurable
Workspace path is hard-coded as /Users/wutianfu/.openclaw/workspace/ making it unusable in other environments. skill-card.md acknowledges this as a risk.
scripts/summarize.sh:4 Make workspace path configurable via environment variable or command argument
Low Sensitive Access
Temporary file creation with PID suffix
Creates /tmp/contacts_$$.txt for temporary storage of extracted contact data before merging
scripts/summarize.sh:46 Consider using mktemp for more secure temporary file handling
Declared capability vs actual capability
Filesystem Pass
Declared READ
→ Inferred WRITE
summarize.sh:63 - cat > "$OUTPUT_FILE" writes to memory file Suspicious artifacts and egress
Medium External URL
https://clawhub.ai/user/gift-is-coding skill-card.md:7
Medium External URL
https://clawhub.ai/gift-is-coding/daily-memory-summary skill-card.md:31
Dependencies and supply chain
There are no structured dependency warnings.
File composition
4 files · 169 lines
Shell 1 files · 84 linesMarkdown 2 files · 80 linesJSON 1 files · 5 lines
Files of concern · 2
scripts/summarize.sh Contact parsing not declared in SKILL.md · Corporate identity extraction with targeted patterns · Hard-coded absolute path not configurable · Temporary file creation with PID suffix
skill-card.md https://clawhub.ai/user/gift-is-coding · https://clawhub.ai/gift-is-coding/daily-memory-summary
Other files · SKILL.md · _meta.json
Security positives
No network requests or external data exfiltration detected
No credential theft or environment variable harvesting
No obfuscation or base64-encoded payloads
No remote code execution or reverse shell capabilities
Core functionality (clipboard aggregation) is legitimate and documented
Skill-card.md provides good risk disclosures including known risks