agent-protocol
PassAudited by VirusTotal on May 11, 2026.
Overview
Type: OpenClaw Skill Name: cs-agent-protocol Version: 1.0.0 The skill bundle defines a structured inter-agent communication protocol for a simulated C-suite team, focusing on coordination, verification, and standardized reporting. It includes robust safety features such as loop prevention (no self-invocation, max depth of 2, circular call detection) and explicit requirements for tagging assumptions and surfacing conflicts. The instructions in SKILL.md and references/invocation-patterns.md are purely logical and aimed at improving the reliability and clarity of multi-agent orchestration without any signs of malicious intent, data exfiltration, or unauthorized execution.
Findings (0)
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.
If used with real company data, information provided to one role may be included in questions, responses, or synthesized outputs involving other roles.
The skill explicitly enables inter-agent communication. This is the core purpose of the skill and is paired with loop-prevention and isolation rules, but users should recognize that business context may be shared across role agents during use.
Any agent can query another using: ``` [INVOKE:role|question] ```
Use this only in contexts where sharing the relevant business information across the selected agent roles is acceptable, and continue to review final outputs for sensitive details.
A poorly framed prompt could influence several role responses at once, producing a broader but still bounded synthesis error.
The skill supports multi-role broadcasts, which can propagate one prompt across several role outputs. The artifacts also include containment controls such as independent responses, depth limits, conflict surfacing, and loop prevention, so this remains purpose-aligned.
CEO can broadcast to all roles simultaneously: ``` [BROADCAST:all|What's the impact if we miss the fundraise?] ```
Use specific, scoped prompts for broadcasts and rely on the skill’s conflict and assumption markers to review uncertain or conflicting outputs.