Skip to content

Claude Skills by galimba

github.com/galimba

3 skillsA× 30 installs4 views

ClarifyAResolve the highest-leverage ambiguities in a draft intake spec before decomposition — one round at a time, a tunable round budget that bounds QUESTIONS, never COVERAGE (F1 route-never-drop). Every answer is logged and propagated; every remaining ambiguity becomes a flagged [ASSUMED]. Interactive asks the human via AskUserQuestion; autonomous never asks. WHAT/WHY, never HOW.ai-agentsgoapiVotes: 0GitHub stars: 4
DecomposeATurn a RATIFIED intake spec (post-Gate-A) into the machine-parseable Task Breakdown — tasks with binary DoD and measurable, tech-agnostic Success Criteria, plus the three machine fields (scope, dod_tests, sc_evidence) the mechanical gate enforces, traced to FR/stories, self-checked via the F8 adversarial coverage pass, then mechanically verified by the Gate-B analyze gate. Emits the spec's task block; does not mint beads.toolsgobashVotes: 0GitHub stars: 4
Spec AuthoringAAuthor an intake spec from templates/spec-template.md — prioritized stories, testable FR-NNN, measurable technology-agnostic SC-NNN — resolving [NEEDS CLARIFICATION] per the Header Mode. Produces the prose spec; hands the Task Breakdown to the decompose skill. WHAT/WHY, never HOW.ai-agentsgoVotes: 0GitHub stars: 4

Claude Skills by galimba — 3 skills, security-graded | Skills Directory