The Agent Skills Directory

[PROMPT_INJECTION]: The skill demonstrates an indirect prompt injection surface through the ingestion of implementation plans. * Ingestion points: Implementation plan files processed by the task-brief script. * Boundary markers: Not explicitly defined in the task briefings. * Capability inventory: Includes file system writes to a local workspace, execution of bundled bash scripts, git operations, and the ability to dispatch subagents with specific instructions. * Sanitization: No explicit content filtering is applied to plan text. The skill includes a mitigation strategy requiring the agent to scan plans for conflicts and escalate ambiguity to the user before starting.
[COMMAND_EXECUTION]: The skill executes several bash scripts (scripts/task-brief, scripts/review-package, scripts/sdd-workspace) to manage its workflow. These scripts use standard unix tools like awk and git. The scripts include safety measures such as set -euo pipefail and input validation via git rev-parse to ensure parameters represent valid git objects before use.

subagent-driven-development