tsukumogami
OrganizationStructured workflow skills for AI coding agents
Categories
Indexed Skills (16)
brief
Structured workflow for creating BRIEF documents — the framing step between a sequenced ROADMAP feature and a PRD's requirements. Use to capture a feature's problem, intended outcome, user journeys, and scope boundary as durable artifacts before requirements are written — including when an issue or conversation already states the problem, since the skill's job is to persist that framing (into the BRIEF, or a downstream PRD/design when a standalone brief is too heavy), not merely to supply framing that is missing. Triggers on "frame this feature", "write a brief for X", "what problem does Y solve before we write the PRD", "we need the framing step between the roadmap and the PRD", or "BRIEF-<name>". Do NOT use for feature sequencing (/roadmap), requirements articulation (/prd), technical architecture (/design), or open-ended exploration (/explore). Drives a six-phase workflow: conversational scoping, structured drafting, structural fill, a two-reviewer jury, and finalization.
charter
Parent skill for the strategic chain. Walks an author through VISION → STRATEGY → ROADMAP as a single conversation, holding state across child boundaries and producing a STRATEGY as the terminal artifact. Use when an author needs strategic framing decided in one sitting rather than reached for one child skill at a time. Triggers on "start a strategic conversation about X", "open a charter for Y", "I need to think through the bet on Z", or direct `/charter <topic>` invocations. Do NOT use when the author already knows which artifact altitude they want (reach for `/vision`, `/strategy`, or `/roadmap` directly).
comp
Structured workflow for creating COMP (competitive-analysis) documents — a private-only artifact that surveys the competitive landscape a feature or product sits in and turns that survey into implications for our own choices. Use when you need to compare competitors along explicit dimensions, find concrete gaps, and connect those findings to decisions, before or alongside writing requirements. Triggers on "competitive analysis for X", "how do competitors handle Y", "survey the market for Z", "what's the competitive landscape", or "COMP-<name>". Do NOT use for feature requirements (/prd), technical architecture (/design), feature framing (/brief), or open-ended exploration (/explore). COMP is private-only: in a public repo the skill refuses and emits a redirect to alternatives.
decision
Structured decision-making skill for contested choices. Use when facing a decision with 3+ viable alternatives, contradicting evidence, or irreversible consequences. Runs a multi-phase evaluation: research, alternative identification, validation bakeoff with adversarial agents, peer revision, cross-examination, and synthesis. Triggers on "decide between X and Y", "which approach should we use for Z", "I need to choose between A, B, and C", or when a lightweight decision escalates via status="escalated". Also invocable as a sub-operation by /design for parallel multi-decision orchestration.
design
Create technical design documents. Use when deciding how to implement something -- the skill decomposes the problem into decision questions, evaluates each with structured trade-off analysis, and produces a design doc with the chosen approach. Triggers on "help me design X", "how should we architect Y", "compare approaches for Z", "write a design doc", "what's the best approach for W", or "I need to decide between A and B". Do NOT use for quick opinions without a formal document, open-ended exploration (/explore), or requirements definition (/prd).
explore
Structured exploration workflow and artifact-type routing advisor. Use when the user isn't sure what to build, doesn't know which workflow fits their situation, or wants to research before committing to a PRD, design doc, or plan. Triggers on "should I write a PRD or a design doc?", "I don't know where to start", "what should I do next?", "how do I start this?", "I'm stuck", or explicit /explore invocations. Helps figure out whether you need a PRD, design doc, plan, or something else through a discover-converge loop with research agents. Does NOT apply when the user already knows their artifact type -- use /prd, /design, or /plan directly instead.
plan
Implementation planning skill. Decomposes a design doc, PRD, roadmap, or directly-stated topic into atomic, sequenced issues with dependency graphs and complexity classifications. Use when given a DESIGN-*.md, PRD-*.md, or ROADMAP-*.md to plan, or when the user says "break this into issues", "plan the implementation", "create issues for this", "decompose this", "what tasks do we need", or "make a plan for X". Also use for direct topic planning without a source document. Produces either a self-contained PLAN doc (single-pr) or GitHub milestone and issues (multi-pr).
prd
Structured workflow for creating Product Requirements Documents (PRDs). Use when writing new PRDs or defining WHAT to build and WHY before implementation. Triggers on "write requirements for X", "define scope for Y", "draft a spec", "what should we build for Z", "I need a PRD", or any request to capture requirements. Do NOT use for technical architecture decisions (/design) or open-ended exploration (/explore). Drives a multi-phase workflow: conversational scoping, parallel research agents, structured drafting, and a 3-agent jury review.
public-content
Content guidelines and restrictions for public repository artifacts -- what to avoid when writing issues, PRs, design docs, and code comments in public repos.
release
Release workflow. Analyzes commits to recommend a version, generates release notes for review, creates a draft GitHub release, dispatches the reusable workflow, and monitors progress. Falls back to draft + manual tag when no workflow is detected.
roadmap
Structured workflow for creating Roadmap documents that sequence multiple features into a coordinated initiative. Use when planning multi-feature work that needs dependency tracking, sequencing rationale, and progress monitoring. Triggers on "create a roadmap for X", "plan the rollout of Y", "sequence these features", or any request to coordinate multiple features into an ordered plan. Do NOT use for single-feature requirements (/prd), strategic justification (/vision), technical architecture (/design), or open-ended exploration (/explore). Drives a multi-phase workflow: conversational scoping, parallel research agents, structured drafting, and jury review.
scope
Parent skill for the tactical chain. Walks an author through BRIEF → PRD → DESIGN → PLAN as a single conversation, holding state across child boundaries and producing a PLAN as the terminal artifact. Use when an author needs feature-scope decided in one sitting rather than reached for one child skill at a time. Triggers on "specify a feature called X", "scope feature Y", "walk me through specifying Z", or direct `/scope <topic>` invocations. Do NOT use when the author already knows which artifact altitude they want (reach for `/brief`, `/prd`, `/design`, or `/plan` directly).
strategy
Structured workflow for creating STRATEGY documents — medium-term defensibility bets that operationalize a piece of an upstream VISION without re-justifying the long-term thesis. Use when the work in front of the team needs a falsifiable bet, building-block decomposition, coordination dependencies, and per-direction invalidation conditions — but doesn't pivot the upstream VISION and isn't ready to commit to a ROADMAP's sequenced feature list. Triggers on "draft a strategy for X", "what's our bet on Y", "operationalize the VISION", "we need a strategic doc between vision and roadmap", or "STRATEGY-<name>". Do NOT use for long-term thesis articulation (/vision), feature sequencing (/roadmap), feature requirements (/prd), or open-ended exploration (/explore). Drives a six-phase workflow: conversational scoping, structured drafting, Building Blocks decomposition, three-reviewer jury, finalization.
vision
Structured workflow for creating Vision documents that capture project thesis, strategic justification, and org fit. Use when defining WHY a project should exist before writing requirements. Triggers on "why should we build X", "define the vision for Y", "justify project Z", "I need a vision doc", or any request to articulate project thesis or strategic positioning. Do NOT use for feature requirements (/prd), technical architecture (/design), or open-ended exploration (/explore). Drives a multi-phase workflow: conversational scoping, parallel research agents, structured drafting, and jury review.
work-on
Implement work end-to-end with branch creation, analysis, coding, tests, and a pull request with CI monitoring. Accepts a GitHub issue (number or URL), a milestone (selects the next unblocked issue), a PLAN document path (drives multiple issues through one shared branch and PR), or a free-form task description. Use when asked to work on, implement, fix, build, tackle, pick up, close, or ship work — at any size, from a single issue to a whole plan.
private-content
Content guidelines for private repository artifacts -- what's allowed when writing issues, PRs, design docs, and code comments in private repos.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.