project-config

Warn

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill frequently uses node -e to execute inline JavaScript snippets for repository reconnaissance, configuration loading, and validation. This pattern of generating and executing code from string templates at runtime is used for cross-platform file operations.
  • [DYNAMIC_EXECUTION]: The helper script scripts/project-config-help.cjs uses dynamic module loading via require() with computed paths to load libraries from the .claude/hooks/lib/ directory.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from docs/project-config.json and .claude/.ck.local.json. This data influences the agent's workflow, including which files are scanned and which capabilities are activated.
  • Ingestion points: docs/project-config.json, .claude/.ck.local.json, and ~/.claude/.ck.json.
  • Boundary markers: No explicit markers are used to isolate configuration data from instructions, although schema validation is applied.
  • Capability inventory: The skill can execute shell commands (node -e), write to the filesystem, and perform repository-wide scans.
  • Sanitization: The skill relies on project-config-schema.cjs for structural validation, but semantic content is interpreted as instructions for the agent's behavior.
  • [COMMAND_EXECUTION]: The skill performs various shell operations for file management and tool invocation, including node, git check-ignore, and references to WinGet for environment repair.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 4, 2026, 10:12 AM