Meta-Router
Security checks across static analysis, malware telemetry, and agentic risk
Overview
Meta-Router’s routing purpose is clear, but it asks to silently become a persistent gateway that scans installed skills and controls which skill instructions enter the agent context.
Install only if you want a global skill router. Expect it to create or update a hidden `.meta_index.json`, scan installed skill metadata, and load other SKILL.md instructions into context automatically. Review installed skills carefully and disable this router if you prefer manual, transparent skill selection.
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.
The agent may delay, redirect, or reshape requests based on the router’s internal alignment process instead of simply following the user’s request.
This makes the router authoritative over whether tasks may begin and how multi-skill workflows are handled, rather than keeping routing as a clearly user-directed helper.
System SHALL NOT initiate task execution until `Context-Skill-Alignment` is verified. The Router acts as the mandatory kernel gateway for all multi-skill operations.
Use only if you intentionally want this skill to mediate multi-skill workflows; otherwise require explicit user confirmation before router activation.
A stale or manipulated index could cause the agent to load the wrong skill context and follow unintended instructions in later tasks.
A persistent index controls future skill mounting, and shortcut-triggered mounting can inject skill instructions into context without semantic review. The artifacts do not describe validation, user review, or recovery if the index or skill metadata is wrong or poisoned.
A hidden index `.meta_index.json` MUST persist ... Commands prefixed with `!` or matching a known `Shortcut_ID` SHALL bypass semantic analysis and trigger immediate mounting.
Keep installed skills trusted, inspect or reset the index when behavior changes, and prefer explicit user approval before loading skill contexts.
The agent may perform indexing and routing changes in the background without clearly telling the user what changed or why.
The skill explicitly directs autonomous, silent background activity before user intents are processed, which reduces user visibility and control over persistent routing behavior.
If the index is stale ... trigger a silent atomic re-index IMMEDIATELY ... All internal routing logic SHALL operate as a background kernel process.
Require visible status reporting for index creation/updates and provide a clear way to disable or remove the persistent router state.