The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The references/setup.md file provides an installation command that fetches a remote script and executes it directly using the Python interpreter: python3 -c "$(curl -fsSL https://raw.githubusercontent.com/yusufkaraaslan/Skill_Seekers/main/setup.py)". This 'curl to interpreter' pattern allows for the execution of unverified remote code.\n- [PRIVILEGE_ESCALATION]: The scripts/enhance-workaround.sh file contains a command that invokes the claude CLI with the --dangerously-skip-permissions flag. This flag explicitly bypasses the agent's built-in security and permission controls, allowing potentially unsafe actions to be performed without user oversight.\n- [COMMAND_EXECUTION]: The skill uses multiple high-privilege shell commands to perform its operations. For example, it uses git clone to fetch external repositories, cp -r to move files into system-like paths (~/.agents/skills-db/), and rm -rf for directory cleanup. These tools are leveraged for automated skill ingestion and setup.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from untrusted external sources, including documentation websites, third-party GitHub repositories, and PDF files. Although it includes a multi-scanner security pipeline, the combination of processing untrusted content and having access to shell execution and safety bypasses creates a significant attack surface.\n
Ingestion points: Workflow 1 (intake_acquire) and Workflow 3 (extract_source) download content from arbitrary user-provided URLs and repositories.\n
Boundary markers: The semantic scanner is configured to run in a 'sandboxed read-only subagent' to limit its impact.\n
Capability inventory: The skill utilizes bash, python3, git, and the claude CLI with administrative-level flags.\n
Sanitization: The skill employs a 3-scanner security funnel (structural, semantic, and AgentShield) to mitigate risks from external content.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates the download of third-party skills and documentation from external GitHub repositories and websites. One such source is raw.githubusercontent.com/yusufkaraaslan/Skill_Seekers/main/setup.py, which is used for dependency installation.\n- [DYNAMIC_EXECUTION]: The skill uses dynamic execution techniques to handle its logic. The bin/flow utility uses subprocess.run to call an inline Python shim for YAML processing, and scripts/enhance-workaround.sh dynamically generates and executes Python code via a heredoc.

skill-builder