coordinating-peer-sessions
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a structured communication protocol using built-in platform tools (
ListAgents,SendMessage) and standard development tools (Git,Linear). It does not download external code or scripts at runtime. - [INDIRECT_PROMPT_INJECTION]: The skill contains explicit defensive instructions to mitigate risks from processing data from other sessions. It directs the agent to treat peer messages as non-authoritative information, specifically prohibiting the execution of slash commands found in messages, preventing permission laundering, and ensuring peer messages are never treated as user consent.
- [COMMAND_EXECUTION]: Tool usage is restricted to local Bash scripts located within the plugin's root directory (
change-provenance.sh,peer-ledger.sh,repo-expert.sh) and standard version control operations. These scripts are used for non-destructive tasks like determining file authorship and managing a local ledger of session claims.
Audit Metadata