synthesis-skills-manager

Pass

Audited by Gen Agent Trust Hub on Apr 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches skill definitions and metadata from the vendor's GitHub repositories (such as github.com/rajivpant/synthesis-skills) to facilitate local installation and updates.
  • [COMMAND_EXECUTION]: Performs file system operations within agent-specific directories, such as ~/.claude/skills/, to manage the lifecycle of methodology skills including installation and state tracking.
  • [PROMPT_INJECTION]: Provides a management interface for ingesting external instructions from remote repositories, which serves as a potential surface for indirect prompt injection.
  • Ingestion points: Processes SKILL.md files retrieved from configured public, private, or shared Git repositories.
  • Boundary markers: Relies on YAML frontmatter structure but does not specify explicit delimiters for the instructional body during the synthesis process.
  • Capability inventory: Possesses the capability to write and modify the agent's active skill set in local configuration directories.
  • Sanitization: Employs a methodology-aware merge process and provenance tracking (via .source.json) to validate and reconcile changes before installation.
  • [COMMAND_EXECUTION]: Implements a dynamic 'synthesis merge' logic that generates updated instruction sets by reconciling differences between local customizations and upstream source versions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 12, 2026, 12:22 AM