The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the Beads CLI tool from established software registries.
References the @anthropic/beads package on the NPM registry for global installation.
Provides configuration for the mise tool to download multi-architecture binaries directly from GitHub releases.
Supports source-based installation via the Go toolchain targeting the steveyegge/beads repository.
[COMMAND_EXECUTION]: Defines a robust set of automation scripts and agent workflows for interacting with the issue tracker.
templates/mise.toml contains numerous Nushell scripts that wrap the bd CLI to manage task lifecycles, dependencies, and sync operations.
agents/beads-worker.md provides a structured workflow for an agent to claim, execute, and close tasks using shell commands and git operations.
[PROMPT_INJECTION]: Contains a vulnerability surface for indirect prompt injection within the automated worker agent.
Ingestion points: The beads-worker agent reads external data from the issue tracker (titles, descriptions, and labels) via the bd show <task_id> --json command.
Boundary markers: There are no explicit delimiters or instructions to treat task-provided content as data rather than instructions.
Capability inventory: The worker agent has extensive permissions including file system modification (Write, Edit), shell execution (Bash), and the ability to delegate to subagents.
Sanitization: The skill does not implement sanitization or validation of the task descriptions before they are used to determine the agent's implementation steps.

beads