bsx

Plugin bsx from marketplace bandersnatch0x-skills in Bandersnatch0x/agent-skills. Commands read from the repo's manifest — 1 month ago.

Skills (16)

• cut-tickets

  Slice a spec into independently verifiable tickets with blocking edges, grill the user on granularity and dependencies until approved, then publish them to the tracker. Use when a spec issue exists and the work must become executable tickets.

• distill-project-manual

  Use when the user asks to create an agent operating manual, handbook, onboarding doc, or agent-facing wiki from an existing repo, or to distill repository docs and conventions into durable agent documentation. Not for general code exploration or human-facing repo summaries.

• draft-spec

  Turn a settled conversation consensus into a spec issue — problem, solution, user stories, implementation and testing decisions, review focus, global constraints — without interviewing the user. Use when the open questions on a topic are answered and the consensus must become a durable spec others c

• implementation-quality-audit

  Audit a messy, agent-built, or half-finished codebase for implementation quality — map → contract checklist → five-axis review → prioritized report. Use before calling agent-built work done, when a repo feels chaotic, or before merge/deploy. Not for single-file nits, style-only review, or a known bu

• timeline-show

  一次长任务的可复核台账。自动开工并按执行追加事件，把 TODO / DOING / DONE、决策点、卡点、检查证据渲染成可双击离线打开、10 秒自刷新的自包含 HTML。用于预计超过 30 分钟的长任务，或在你说「看全貌 / timeline-show」时查看当前全貌。

• visualize

  Render a repository's agent workflow and its tracker state as self-contained HTML+SVG views — a static overview of pipeline stages and cross-cutting mechanisms read from the repository's own workflow document, and a frozen census board of a map's sub-issue state with blocking edges. Use when someone

• close-functional-loop

  use this when work piled up without a closable end-to-end loop: missing done-criteria, fuzzy module boundaries, or partial features that block progress; run before coding or agent swarms

• feature-scope-triage

  use this when a project has accumulated too many features and you need a keep / cut / defer / spin-out decision per feature against one stated positioning

• goals-and-success-metrics

  use this when project goals feel vague and you need a North Star, supporting metrics, stage milestones, and anti-goals to stop feature sprawl

• idea-stress-test

  Stress-test the idea a repository is built on into an investment decision — falsifiable hypotheses, evidence for and against from the repo and the outside world, the single bottleneck belief most likely to change the decision, and the highest-information next experiment. Use when the user asks wheth

• product-positioning-audit

  use this when a project's positioning feels unclear or drifts from what the repo actually builds and you need an evidence-based audit of who it is for, what job it does, and what it is NOT

• product-scope-audit

  Evidence-backed continue / narrow / reposition / pause-to-validate / stop recommendation for an early or drifting product, plus one next action — auditing capability boundaries (own / integrate / defer / exclude), end-to-end journeys, and broken handoffs to find the smallest coherent product worth c

• target-user-and-jtbd

  use this when who the product is really for is unclear and the real jobs-to-be-done must be pinned down with evidence before positioning or cutting features

• claude-retrospective

  Distill evidence from Claude Code sessions into one minimal durable behavior change. Use when reviewing a time, project, or failure window; investigating repeated corrections or friction; or deciding whether to update instructions, memory, or a reusable skill.

• claude-session-manager

  Maintain Claude Code session state safely. Use when diagnosing slow or large local state, preparing a resumable handoff, or planning or executing archival and cleanup of sessions, worktrees, logs, or project state.

• find-hackathons

  Find and verify currently enterable hackathons against hard constraints — deadline, location, eligibility, tracks, required technology, prizes — then optionally generate a bilingual local filterable webpage with participant-profile matching and agent-ready brainstorm briefs. Use when the user wants

Similar plugins