Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Openclaw phone

v1.0.1

Use CallMyCall API to start, end, and check AI phone calls, and return results in chat. Use when the user asks to call someone, plan a future call, end a cal...

1· 426·0 current·0 all-time
byBenjamin Waye@benjaminwaye
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
!
Purpose & Capability
The SKILL.md and accompanying files clearly require a CALLMYCALL_API_KEY and explain how it's used; however the registry metadata at the top of the report lists no primary credential and 'Required env vars: none'. That mismatch is an incoherence — the skill will need an API key but the registry entry doesn't declare it, which can lead to missing prompts/validation in the platform. Other than that, the actions described (start/end/list calls, verify caller IDs, fetch recordings/transcripts) are coherent with a CallMyCall phone skill.
Instruction Scope
SKILL.md stays narrowly scoped to the call-management purpose: gathering phone, language, brief; doing validation; calling the CallMyCall API; storing a small recent_calls state. It explicitly forbids creating OS schedulers, storing API keys in skill files/state, and autonomous background runs. Two items worth noting: (1) the docs recommend using a specific backend base URL (https://call-my-call-backend.fly.dev) in addition to api.callmycall.com — using a non-official domain should be validated; and (2) the API accepts highly sensitive fields (e.g., personal_security_number inside a persona object) — the instructions do not forbid including PII in requests, so agents could transmit sensitive user data if prompted to do so.
Install Mechanism
This is an instruction-only skill with no install spec and no code files to execute at install time, so there is no downloader/extractor or package install risk.
!
Credentials
The skill legitimately needs a single service credential (CALLMYCALL_API_KEY) but the registry metadata does not declare it (incoherent). The SKILL.md correctly describes a limited key resolution order (env var, user config, prompt) and forbids persisting keys automatically, which is good. However the API surface allows sending PII and webhook URLs; ensure any API use is deliberate and avoid entering persistent credentials unless you trust the service and skill owner. The recommendation to prefer a fly.dev backend (different from the public api.callmycall.com hostname) raises an extra review step: confirm the correct base URL for your account.
Persistence & Privilege
The skill does not request always:true, does not ask to modify other skills, and explicitly tells agents not to create background schedulers or persist credentials automatically. It stores minimal per-skill state (recent_calls) which is appropriate for the described functionality.
What to consider before installing
Before installing: (1) confirm the platform/registry metadata is updated to declare CALLMYCALL_API_KEY — the skill expects that key even though the top-level registry entry did not list it; (2) verify the correct CallMyCall base URL for your account (the docs mention both api.callmycall.com and a fly.dev backend — prefer the official domain or confirm the fly.dev host is legitimate for the provider); (3) avoid storing long-lived API keys in the skill — follow the SKILL.md advice to use an env var or one-time interactive key and only persist manually if you understand the risk; (4) be cautious about sending sensitive personal data (the API supports fields like personal_security_number in a persona object) — do not include PII unless absolutely necessary and authorized; (5) ask the skill author/owner (or registry operator) to fix the registry metadata mismatch so the platform can surface the credential requirement correctly. These issues suggest sloppy/unfinished packaging rather than explicit malice, but you should validate ownership and endpoints before granting credentials.

Like a lobster shell, security has layers — review code before you run it.

latestvk979hf2fg8h2n1c90t6yby6b6x82pn5r

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments