Executes self-contained tasks in an isolated forked context and returns a compact summary.
npx skills add https://github.com/vedang/dotagents --skill boomerang-vedang
Please help me install this Agent Skill. Skill: boomerang Source: https://github.com/vedang/dotagents/tree/main/claude/skills/boomerang Command: npx skills add https://github.com/vedang/dotagents --skill boomerang-vedang
High-intent search queries and answers about installing and using this skill.
Invoke the boomerang skill with a self-contained task description. It forks a separate context where a general-purpose agent completes the work, and only a compact summary of the outcome returns to your main conversation.
Self-contained tasks whose intermediate steps do not matter to the ongoing conversation work best, such as bulk renames, dependency updates, or test fixes. Tasks needing user decisions mid-flight are poor fits since the forked agent cannot ask questions.
No, the forked agent explicitly does not commit. It follows global instructions like hygiene and quality gates but leaves committing to the main conversation, listing follow-ups in its summary.
The agent marks the outcome as partial or blocked in its summary and explains why in one sentence. It also lists assumptions made and follow-ups the main agent or user still needs to handle.
Avoid it when the task requires interactive clarification, when you need to observe intermediate reasoning, or when the work is tightly coupled to decisions being made in the main conversation.
ste-bahGenerate a 14-section Product Requirements Document from a project description.
doravidanReview RALPH worktree diffs, commits, and file changes before merging to main.
dorrioCalculate a Quality Score and gate verdict for software development stories.
marktantongcoProvide numbered step-by-step reasoning with assumptions and confidence assessments.
