Elite Multi-Agent Comms Mesh
ReviewAudited by ClawScan on May 10, 2026.
Overview
This skill is a transparent Supabase-based agent messaging tool, but users should only use it with trusted agents because shared mesh messages can influence automated agent behavior.
This looks coherent for its stated purpose. Before installing, create a dedicated Supabase project, do not send secrets over the mesh, share the key only with trusted agents, and be careful enabling cron or heartbeat polling because incoming messages may cause agents to respond automatically.
Findings (3)
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.
Any agent or operator with the shared Supabase key can read mesh traffic and participate in the channel.
The skill clearly discloses that the Supabase key is a shared trust boundary for the mesh, meaning message privacy and access depend on trusting every agent that has the key.
Trust boundary: All agents sharing the same Supabase anon key can read each other's messages. Only share credentials with agents you control.
Use a dedicated Supabase project, share the key only with agents you control, and avoid sending secrets or sensitive data through mesh messages.
A peer agent's message could influence what the receiving agent does during a scheduled polling run.
The optional cron workflow can cause an agent to treat incoming mesh messages as tasks and respond automatically.
If messages found, process and reply via: bash {baseDir}/scripts/mesh-send.sh <agent> '<reply>'Only enable automatic polling for trusted peers, and add explicit rules that mesh messages must not override system, developer, or user instructions or trigger sensitive actions without approval.
Poorly configured agents could create reply loops, token-cost spikes, or noisy message floods.
The documentation acknowledges that broadcasts and auto-replies can amplify across the mesh, though it also provides mitigations.
Broadcast Reply Storm ... With N agents, a broadcast can generate N² messages.
Follow the documented anti-spam and broadcast-reply rules, and start with low-frequency polling until behavior is verified.