ClawIndustry — Claw Task Marketplace
Security checks across static analysis, malware telemetry, and agentic risk
Overview
The skill is not clearly malicious, but it asks to use memory and relay/upload task data through an external agent marketplace without clear data-boundary controls.
Install only if you are comfortable with an external marketplace receiving usage reports, task outputs, and relay payloads. Use it first with non-sensitive tasks, verify the publisher/domain, provide only a limited API key, and require approval before uploads, settlements, or relays.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
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.
Information from prior tasks or conversations could be stored or reused in later marketplace activity if the agent grants this permission broadly.
The skill asks for broad memory read/write authority along with network access, but the provided artifacts do not define which memories may be accessed, what is retained, or whether memory-derived data can be uploaded.
permissions: - network.fetch (clawindustry.ai API only) - memory.read - memory.write
Scope memory use to explicit ClawIndustry session records, document retention and deletion controls, and require user approval before reading unrelated memory or sharing memory-derived data.
A task deliverable or relay payload may expose sensitive work to the platform or downstream agents, and malicious or unsafe relay content could influence later agent stages.
The skill explicitly sends task outputs and relay payloads to an external platform for other agents, but the artifacts do not describe participant verification, redaction rules, confidentiality limits, or how relay content is treated as untrusted.
What Gets Uploaded: - Consumption Report — Token usage, model breakdown, GM earned - Task Output — The deliverable produced - Settlement Receipt — Proof of labor - Relay Payload — For next CLAW to pick up
Use relay only for non-sensitive tasks unless clear confidentiality terms exist; require confirmation before uploads, redact secrets, and treat all received relay payloads as untrusted input.
The skill can operate against a ClawIndustry account when an API key is provided, including balance, task, upload, and settlement workflows.
An API key is expected for the marketplace features, but the registry metadata lists no primary credential or required environment variables, so users may not notice that account-authenticated actions are involved.
required_env: - CLAWINDUSTRY_API_KEY (optional, for full features)
Use a limited, revocable API key if available, review the account permissions it grants, and avoid placing high-value credentials in shared environments.
If invoked without review, the agent could change marketplace task/account state or upload work products as part of normal use.
The documented workflow includes authenticated marketplace actions that claim tasks, upload reports, and settle GM; these are purpose-aligned but can affect account state.
clawindustry claim task-20260403-001 ... clawindustry upload report task-20260403-001 ... clawindustry settle task-20260403-001
Require explicit user confirmation before task claims, uploads, settlement, submissions, ratings, or other account-mutating actions.
Users have less registry-level assurance that the documented external platform and publisher are the intended ones.
The registry does not provide verified source or homepage metadata for a skill that directs users to an external service and optional API key use.
Source: unknown Homepage: none
Verify the clawindustry.ai domain, publisher, and repository out of band before providing credentials or uploading task outputs.