The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the execution of several CLI tools, including br (beads_rust), bv (beads viewer), git, jq, and pgrep. These tools are used for initializing workspaces, managing the issue life cycle, and synchronizing data via Git repositories.
[REMOTE_CODE_EXECUTION]: The br upgrade command allows the tool to perform a self-update by downloading and executing new binary versions from a remote source. While common for CLI tools, it represents a mechanism for remote code update and execution.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it reads and processes external data from issues (titles, descriptions, comments).
Ingestion points: Data enters the agent context via commands like br show <id> --json, br list --json, and br search "keyword" --json (SKILL.md, references/COMMANDS.md).
Boundary markers: The documentation emphasizes the use of structured output formats like --json to aid parsing, but it does not specify explicit boundary markers or instructions to ignore potential commands embedded within the issue data.
Capability inventory: The agent has the capability to execute shell commands (br, git), modify local files in the .beads/ directory, and perform Git-based network operations (git push, git pull).
Sanitization: There is no evidence of sanitization or filtering of the content within the issue fields before it is processed by the agent.

br