The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The workflow ingests implementation plans and specifications provided as external files. These inputs are used to generate task briefs for implementer subagents and context for reviewer subagents. A malicious plan could attempt to influence subagent behavior through instructions embedded in the task descriptions.
Ingestion points: The skill reads external PLAN_FILE and referenced specifications at setup, and uses scripts/task-brief to extract requirements.
Boundary markers: The skill mandates the use of fresh subagents with isolated contexts and dedicated workspaces for each task to prevent context leakage across the session.
Capability inventory: The skill possesses capabilities to execute local scripts, perform file system operations, and run git commands (including destructive ones).
Sanitization: The skill does not describe any explicit sanitization or validation of the content extracted from plan files before it is passed to subagents.
[COMMAND_EXECUTION]: The skill utilizes several local utility scripts (scripts/sdd-workspace, scripts/task-brief, scripts/review-package) and standard git operations (git rev-parse, git log, git diff, git clean -fdx) for workspace management, task extraction, and review package generation. The git clean command is specifically mentioned for workspace teardown.

subagent-driven-development