The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to perform global installations of NPM packages (ecc-universal and ecc-agentshield) from an unverified third-party GitHub repository (affaan-m/everything-claude-code). This practice introduces a supply chain risk, as the integrity of code from non-official or unvetted sources cannot be guaranteed.
[INDIRECT_PROMPT_INJECTION]: The workflow incorporates the ingestion of external data via tools like gh search and agent-reach without explicit security boundaries.
Ingestion points: Data is retrieved from GitHub search results and external web sources during the Research & Reuse phase (SKILL.md).
Boundary markers: No delimiters or instructions are provided to ensure the agent ignores potentially adversarial content within the fetched data.
Capability inventory: The agent has the ability to execute shell commands, perform package installations, and interact with git repositories (SKILL.md).
Sanitization: The skill lacks mentioned procedures for sanitizing or validating external content before it influences agent decision-making or code generation.
[COMMAND_EXECUTION]: The skill prescribes several shell commands for environment setup and operational tasks, including npm install -g, git diff --cached, and gh search.

ecc-workflow