Flux

Security checks across malware telemetry and agentic risk

Overview

Flux appears to be a legitimate shared-state integration, but it stores agent data persistently and can use powerful tokens for deletion or admin changes.

Install this only if you intend agents to coordinate through Flux. Use a namespace-scoped FLUX_TOKEN, avoid setting FLUX_ADMIN_TOKEN unless you need admin changes, confirm delete or batch-delete actions before running them, and do not publish secrets or sensitive private data to shared Flux state.

VirusTotal

66/66 vendors flagged this skill as clean.

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.

#
ASI04: Agentic Supply Chain Vulnerabilities
Info
What this means

The skill may not work unless curl is available, and users should understand it makes HTTP requests to the configured Flux service.

Why it was flagged

The skill’s own documentation says curl is required, while the registry requirements list no required binaries. This is an under-declared but disclosed setup dependency, not hidden behavior.

Skill content
2. **curl** installed (required)
Recommendation

Declare curl as a required binary in metadata and review the included script before using it.

#
ASI02: Tool Misuse and Exploitation
Medium
What this means

If used carelessly, an agent could remove shared Flux entities or change runtime settings on the configured Flux instance.

Why it was flagged

The CLI exposes batch deletion of state entities and admin configuration updates. These are disclosed management features, but they are higher-impact than simple publish/query operations.

Skill content
batch-delete) ... api_call POST "/api/state/entities/delete" "$filter" ... admin-config) ... curl -s -X PUT "${FLUX_URL}/api/admin/config"
Recommendation

Require explicit user confirmation for delete, batch-delete, and admin-config updates; use narrow namespace filters and least-privilege tokens.

#
ASI03: Identity and Privilege Abuse
Medium
What this means

A session with FLUX_ADMIN_TOKEN can update Flux runtime configuration, and FLUX_TOKEN authorizes access to the configured Flux namespace or instance.

Why it was flagged

The skill requires a bearer token and optionally supports an admin token. This is expected for the Flux API integration, but the admin token grants broader authority.

Skill content
"requires": { "env": ["FLUX_TOKEN"] }, "primaryEnv": "FLUX_TOKEN", "optionalEnv": ["FLUX_URL", "FLUX_ADMIN_TOKEN"]
Recommendation

Use least-privilege Flux tokens, store them securely, and only set FLUX_ADMIN_TOKEN in sessions that truly need admin configuration changes.

#
ASI06: Memory and Context Poisoning
Medium
What this means

Information published to Flux may persist and influence other agents later; incorrect or malicious observations could be treated as canonical state.

Why it was flagged

The core design is persistent shared state reused by agents. This is disclosed and purpose-aligned, but shared memory can carry sensitive data or untrusted observations forward.

Skill content
Flux is a persistent, shared, event-sourced world state engine. Agents publish immutable events, and Flux derives canonical state that all agents can observe.
Recommendation

Do not publish secrets or private data unless intended, use namespace isolation, validate important state before acting on it, and define cleanup or retention practices.