The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The scripts/install-dap.sh script downloads a binary executable from github.com/AlmogBaku/debug-skill at runtime.
[REMOTE_CODE_EXECUTION]: The skill facilitates the download and execution of the dap binary. While the installer script scripts/install-dap.sh includes SHA-256 verification, the binary itself originates from an unverified community repository.
[COMMAND_EXECUTION]: The skill relies on executing various CLI tools including dap, brew, go, and language-specific debug adapters like debugpy and dlv.
[PRIVILEGE_ESCALATION]: The documentation in references/installing-debuggers.md suggests using sudo for granting debugging permissions (DevToolsSecurity) and installing system packages (apt install).
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by ingesting and displaying live program state.
Ingestion points: Untrusted data enters the agent context via dap debug, dap eval, and dap inspect (file: SKILL.md).
Boundary markers: The skill lacks explicit delimiters or warnings to ignore instructions embedded within the debugged program's state (file: SKILL.md).
Capability inventory: The agent can execute shell commands via the dap tool and interact with the local file system (file: SKILL.md).
Sanitization: There is no evidence of sanitization or filtering of external variable contents before presentation to the agent (file: SKILL.md).
[DYNAMIC_EXECUTION]: The skill enables dynamic code evaluation through dap eval (file: SKILL.md), allowing the agent to execute arbitrary expressions against a running process context.

debugging-code