sync-skills-shared-protocols

Warn

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill's core workflow involves executing several local Python and Node.js scripts (e.g., sync-update-blocks.py, build-protocol-projection.cjs, sync-hooks-to-skills.py) to automate file modifications. These scripts are invoked via shell commands and accept user-provided tags as arguments.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the propagation of instruction content from a central canonical source (.claude/skills/shared/sync-inline-versions.md) to a large number of agent and skill files. This architectural pattern creates an ingestion surface where changes to the source file are automatically distributed across the agent ecosystem, potentially allowing for the spread of malicious instructions if the source is manipulated.
  • [PRIVILEGE_ESCALATION]: The skill documentation explicitly grants authorization for the spawn_agent capability when executing its protocols. This allows the skill to initiate sub-agents, which increases the complexity and reach of the operations performed under the user's initial authorization.
  • [DYNAMIC_EXECUTION]: The skill utilizes scripts to dynamically generate and inject "mirrors", "projections", and routing metadata (inject_agent_skill_connections.py) into agent instruction files. This involves programmatically assembling and writing executable agent context at runtime.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 4, 2026, 10:11 AM