The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill contains commands that fetch shell scripts from the internet and pipe them directly into a bash shell. This allows for arbitrary code execution from unverified sources.\n
Evidence: assets/beads-backend.yml contains install: "curl -fsSL https://raw.githubusercontent.com/steveyegge/beads/main/install.sh | bash".\n
Evidence: assets/beads-viewer.yml contains install: "curl -fsSL https://raw.githubusercontent.com/Dicklesworthstone/beads_viewer/main/install.sh | bash".\n- [EXTERNAL_DOWNLOADS]: Installation instructions depend on downloading content from personal GitHub repositories that are not associated with trusted organizations or well-known technology service providers.\n- [COMMAND_EXECUTION]: The skill operations involve executing various shell commands and specialized CLI tools (gh, bd, bv, ls, grep, awk) to manage tasks and process files.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it retrieves and processes untrusted data from external backends (titles, descriptions, comments) and integrates it into the agent context.\n
Ingestion points: Task data is ingested via tool calls like gh issue list, bd list, and Linear MCP resources in assets/github-backend.yml, assets/beads-backend.yml, and assets/linear-backend.yml.\n
Boundary markers: There are no delimiters or instructions to ignore commands potentially embedded in the task content retrieved from backends.\n
Capability inventory: The skill has access to shell execution and file system manipulation tools.\n
Sanitization: No sanitization or escaping is applied to the retrieved task descriptions before they are used in commands or logs.

backlog-manager