memory-sync-setup
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The generator uses a local shell script (
scripts/generate.sh) to perform repository analysis and file management tasks, including the use ofgit ls-files,find, andawkfor placeholder substitution and metadata management. - [DYNAMIC_EXECUTION]: The skill functions as a generator, producing a project-specific
/memory-synctool by populating templates with derived repository metadata and AI-generated fact tables. - [INDIRECT_PROMPT_INJECTION]: The generated skill processes repository contents (such as
CLAUDE.mdand rule files) through subagents that verify facts against the codebase. - Ingestion points: Repository files (rules, CLAUDE.md, agents, skills) are read by synchronization agents to identify and verify facts.
- Boundary markers: Batch prompts define the scope and role for subagents but do not implement strict delimiters for wrapping untrusted repository content.
- Capability inventory: Subagents are granted access to
Read,Edit,Bash, andAgenttools. - Sanitization: Content is checked for domain relevance and fact-verified against the codebase, with a mandate to purge obvious or generic knowledge.
Audit Metadata