Claude Skills by uHappyLogic

Autonomously sweep the current milestone's requirements for open and deferred questions that already carry an embedded recommendation and record each one's recommendation as its answer — by dispatching the file-editing answer-open-question-with-recommendation agent once per question, strictly sequentially, and committing each answer itself after the agent returns. Use this after a recommend sweep has annotated the questions (via /recommend-all-open-questions) and you want every recommendation...

Record a chosen `<alternative>` from an open question's embedded analysis as its answer in the current milestone's requirements.md inline in this conversation, then commit that answer on its own. Use when the recommend sweep has annotated a question with alternatives and you want to accept a specific alternative — named by its id, not necessarily the recommended one — as the decision for a single named question you want to stay available to discuss afterwards.

Record one open question's embedded recommendation as its answer in the current milestone's requirements.md inline in this conversation, then commit that answer on its own. Use when the recommend sweep has annotated a question with a recommendation and you want to accept and record that recommendation as the decision for a single named question you want to stay available to discuss afterwards.

Answer a named open question in the current milestone requirements document, recording the decision and its downstream implications and committing the manual-answer edit.

Use when the user is asking a question about the current milestone rather than trying to change it.

Distill reusable answering principles from a just-finished milestone's recorded decisions into the project-wide principle store. This is the optional finish-time follow-up that /finish-current-milestone recommends — run it right after finishing a milestone to harvest everything that milestone's manual answers taught. It walks the milestone's `Manual-answer:` commits, extracts the generalizable keep/eliminate rules behind them, dedups the candidates against each other, then confirms each with ...

Complete all tasks from the current milestone's TASKS_TODO.md one by one, committing after each completed task.

Complete a single named task from the current milestone's TASKS_TODO.md inline in this conversation, then move it to TASKS_DONE.md. Use for one ad-hoc task you want to stay available to discuss and tweak afterwards.

Create a new milestone directory with initialized requirements.md (Goal filled from the provided description) and empty TASKS files.

Convert the current milestone's requirements.md into a complete, dependency-ordered TASKS_TODO.md. Decomposes the milestone into high-level task briefs, proves every requirement is covered, then writes those briefs directly into the task list as brief-level task sections.

Analyze an overall goal description and ask clarifying questions to make the milestone goal crisp, complete, and ready for formal definition.

Clarify a rough or ambiguous issue surfaced mid-flight into one or more clear tasks, then hand off to /submit-task. Use this whenever the user reports a problem, gap, or "we should also..." idea mid-flight but the description is too vague to act on, or whenever they want to talk through an issue before adding it to the milestone task list. Also use it when the user flags that something is a bigger chunk of work that may need several task entries, or when a single reported issue turns out to b...

Start a structured conversation about a named open question in the current milestone requirements — surfaces alternatives, trade-offs, and a recommendation to help the user reach a decision.

Mark the current milestone as done — records accomplishments in milestones/README.md, clears the current-milestone pointer to "none" in milestones/README.md, and updates CLAUDE.md only for lasting changes to the project's environment context.

Activate an already-defined milestone — scans milestones/ for a defined-but-not-yet-active directory and updates milestones/README.md to point to it as the current milestone. Requires finish-current-milestone to have been run first (current pointer must be "none").

One-time bootstrap for the milestone workflow in a project — creates the milestones/ directory and milestones/README.md with the Current Milestone pointer, and ensures CLAUDE.md carries the Milestone Workflow guidance. Safe to run on an existing project; never overwrites existing files.

Revise the `## Goal` of the current milestone's requirements.md when a discussion reveals the goal itself — not just an open question — needs to change. Use when the user wants to reshape, broaden, narrow, or correct the objective of an already-defined milestone (e.g. "the goal should also cover X", "drop Y from the goal", "the goal is really about Z"). This is the only skill that mutates an existing milestone's Goal; it edits the Goal section only and surfaces — never auto-cascades — the dow...

Non-interactively sweep the current milestone's requirements for open and deferred questions and annotate each one with an embedded recommendation — alternatives plus a single recommended option — produced by dispatching one read-only recommend-open-question subagent per question. Use this to batch the per-question /discuss-open-question deliberation across every question at once, typically right after a /review-milestone-requirements pass, so the questions arrive at /answer-open-question-wit...

Run a review pass over the current milestone's requirements.md to drive its open-questions loop forward — reconcile the existing question set against what's already been decided (prune blocks an existing decision now covers, dedup repeats), surface genuinely new gaps the latest decisions exposed, and report whether the document has converged enough to derive tasks. This is the repeatable engine of the \"iterate the requirements\" loop: run it after define/specify to open the questions, and re...

Analyze the project's existing state and fill the "Relevant starting state" section of a milestone's requirements.md with context that supports future decisions.

Add a single issue (surfaced during development) to the current milestone's TASKS_TODO.md as a brief-level task section. Use when the user reports a concrete bug, gap, or "we should also..." that is already clear enough to queue. Triages for duplicates, decides where the task belongs, then authors and inserts the task inline.

Autonomously sweep the current milestone's requirements for open and deferred questions that already carry an embedded recommendation and record each one's recommendation as its answer — by dispatching the file-editing answer-open-question-with-recommendation agent once per question, strictly sequentially, and committing each answer itself after the agent returns. Use this after a recommend sweep has annotated the questions (via /recommend-all-open-questions) and you want every recommendation...

Record a chosen `<alternative>` from an open question's embedded analysis as its answer in the current milestone's requirements.md inline in this conversation, then commit that answer on its own. Use when the recommend sweep has annotated a question with alternatives and you want to accept a specific alternative — named by its id, not necessarily the recommended one — as the decision for a single named question you want to stay available to discuss afterwards.

Record one open question's embedded recommendation as its answer in the current milestone's requirements.md inline in this conversation, then commit that answer on its own. Use when the recommend sweep has annotated a question with a recommendation and you want to accept and record that recommendation as the decision for a single named question you want to stay available to discuss afterwards.

Answer a named open question in the current milestone requirements document, recording the decision and its downstream implications and committing the manual-answer edit.

Use when the user is asking a question about the current milestone rather than trying to change it.

Distill reusable answering principles from a just-finished milestone's recorded decisions into the project-wide principle store. This is the optional finish-time follow-up to /finish-current-milestone — run it right after finishing a milestone to harvest everything that milestone's manual answers taught. It walks the milestone's `Manual-answer:` commits, extracts the generalizable keep/eliminate rules behind them, dedups the candidates against each other, then confirms each with you one at a ...

Complete all tasks from the current milestone's TASKS_TODO.md one by one, committing after each completed task.

Complete a single named task from the current milestone's TASKS_TODO.md inline in this conversation, then move it to TASKS_DONE.md. Use for one ad-hoc task you want to stay available to discuss and tweak afterwards.

Create a new milestone directory with initialized requirements.md (Goal filled from the provided description) and empty TASKS files.

Convert the current milestone's requirements.md into a complete, dependency-ordered TASKS_TODO.md. Decomposes the milestone into high-level task briefs, proves every requirement is covered, then writes those briefs directly into the task list as brief-level task sections.

Analyze an overall goal description and ask clarifying questions to make the milestone goal crisp, complete, and ready for formal definition.

Clarify a rough or ambiguous issue surfaced mid-flight into one or more clear tasks, then hand off to /submit-task. Use this whenever the user reports a problem, gap, or "we should also..." idea mid-flight but the description is too vague to act on, or whenever they want to talk through an issue before adding it to the milestone task list. Also use it when the user flags that something is a bigger chunk of work that may need several task entries, or when a single reported issue turns out to b...

Start a structured conversation about a named open question in the current milestone requirements — surfaces alternatives, trade-offs, and a recommendation to help the user reach a decision.

Mark the current milestone as done — records accomplishments in milestones/README.md, clears the current-milestone pointer to "none" in milestones/README.md, and updates CLAUDE.md only for lasting changes to the project's environment context.

Activate an already-defined milestone — scans milestones/ for a defined-but-not-yet-active directory and updates milestones/README.md to point to it as the current milestone. Requires finish-current-milestone to have been run first (current pointer must be "none").

One-time bootstrap for the milestone workflow in a project — creates the milestones/ directory and milestones/README.md with the Current Milestone pointer, and ensures CLAUDE.md carries the Milestone Workflow guidance. Safe to run on an existing project; never overwrites existing files.

Revise the `## Goal` of the current milestone's requirements.md when a discussion reveals the goal itself — not just an open question — needs to change. Use when the user wants to reshape, broaden, narrow, or correct the objective of an already-defined milestone (e.g. "the goal should also cover X", "drop Y from the goal", "the goal is really about Z"). This is the only skill that mutates an existing milestone's Goal; it edits the Goal section only and surfaces — never auto-cascades — the dow...

Non-interactively sweep the current milestone's requirements for open and deferred questions and annotate each one with an embedded recommendation — alternatives plus a single recommended option — produced by dispatching one read-only recommend-open-question subagent per question. Use this to batch the per-question /discuss-open-question deliberation across every question at once, typically right after a /review-milestone-requirements pass, so the questions arrive at /answer-open-question-wit...

Run a review pass over the current milestone's requirements.md to drive its open-questions loop forward — reconcile the existing question set against what's already been decided (prune blocks an existing decision now covers, dedup repeats), surface genuinely new gaps the latest decisions exposed, and report whether the document has converged enough to derive tasks. This is the repeatable engine of the "iterate the requirements" loop: run it after define/specify to open the questions, and re-r...

Analyze the project's existing state and fill the "Relevant starting state" section of a milestone's requirements.md with context that supports future decisions.

Add a single issue (surfaced during development) to the current milestone's TASKS_TODO.md as a brief-level task section. Use when the user reports a concrete bug, gap, or "we should also..." that is already clear enough to queue. Triages for duplicates, decides where the task belongs, then authors and inserts the task inline.