All authors

Claude Skills by Ugbot
github.com/Ugbot79 skills0 installs3 views
- Venus Map RecordHow to write a Venus area MAP.md record (and the MAP.waveN-<packet>.md note that becomes one), newest-first dated section with the ticket key, the three completion states named separately, measured numbers cited to logs, production consumer file:line, open owner questions by doc 33 §6.5 number, remaining known defects. Use when closing a packet, folding wave notes, or when a MAP claim needs correcting.Votes: 0GitHub stars: 9
- Venus Packet PreflightImplementer pre-flight for a Venus wave packet, the checklist and the tools/check_packet.py script that catch the defect classes reviewers rejected in every Stage A wave (bare assert() in Release, engine getenv levers, vacuous WILL_FAIL controls, unmeasured numbers, 70-line functions, comment terminators). Run BEFORE handing off any packet; paste its block into the handoff.Votes: 0GitHub stars: 9
- Venus Render DebugDiagnose Venus rendering bugs (VK-first), logging discipline, validation layers, per-pass GPU timers, visual bisection, and the failure signatures already catalogued (black output, garbage polygons, stripes, Y-flips, descriptor pool errors). Use when a demo renders wrong/black/corrupt, validation errors appear, or perf budgets need checking.Votes: 0GitHub stars: 9
- Venus Smoke GoldenRun and interpret the flagship smoke test + VK golden-image regression, the gating visual check for renderer work. Use when verifying renderer changes, chasing view FAIL/var=0/exit-code failures, recording new goldens after a visual sign-off, or when asked "is the renderer still green".Votes: 0GitHub stars: 9
- Venus TestsRun Venus's test suites the right way, ctest labels, the key gates per subsystem, the 20-second rule, and headless SERVERSIDE testing. Use when verifying changes, deciding which tests gate a diff, or adding a new test target.Votes: 0GitHub stars: 9
- Venus Wyrd ConnectedRun WYRDHOLM connected, build the Linux server / headless-client Docker images from a SHA with a build label, keep Docker disk under control, drive the real dedicated server + headless witnesses + graphical review, run the join probe legs L1-L6, and write the connected receipt. Use for any Stage A/B connected evidence (join, reconnect, encounter, crate, route), never for a Windows server exe.Votes: 0GitHub stars: 9
- Bench Rdtsc ProfileInline-RDTSC micro-profiling in MarbleDB's C++ benchmarks. Use when a YCSB workload (A–F, Load) is slower than expected and you need to know which sub-phase is burning cycles before proposing a fix. Faster than launching PerfView and gives exact cycle-per-op accounting for the specific code path you care about.Votes: 0GitHub stars: 9
- Chukonu DevBuild, test, and fuzz chukonu through one driver (tools/ckdev.py) across both toolchains: Windows MSVC Release (the perf + correctness gate) and the Linux Clang+ASAN Debug container (asserts ON, finds masked bugs). Use when the user asks to build, run tests/ctest, run the seeded fuzzer, replay a fuzz seed, check the TPC-H coverage gate/board, run the ASAN synthetic all-22, or "run the whole loop / everything". The fuzzer is deterministic — every failure prints a seed that replays exactly.Votes: 0GitHub stars: 9
- Debug Linux LldbStep-through debugging chukonu test executables under Linux (in a Docker container) with LLDB and AddressSanitizer. Use when a Windows-only bug needs Linux reproduction, or you need ASAN to catch a heap-OOB / use-after-free the MSVC debugger can't see. Clang/LLD only — never GCC, never mingw.Votes: 0GitHub stars: 9
- Debug Windows MsvcStep-through debugging chukonu test executables with Visual Studio's MSVC debugger on Windows. Use when a test fails locally, an `.exe` segfaults, or you need to watch operator state mid-pipeline. Prefer this over `fprintf` debugging — one breakpoint beats five rebuild-and-print cycles.Votes: 0GitHub stars: 9
- Llm Station AnalyzeAnalyze code structure using LLM Station's call graph, predicate search, and task decomposition. Use when understanding function relationships, impact of changes, or planning implementation order.Votes: 0GitHub stars: 9
- Llm Station DebugDebug and profile code using LLM Station's devtools integration. Run Valgrind memcheck, perf profiling, or check available debugging tools.Votes: 0GitHub stars: 9
- Llm Station PatternsSearch for and instantiate design patterns and reusable code templates. Includes Gang of Four patterns (Observer, Strategy, Factory, etc.) and previously crystallized action templates. Always check before writing new code.Votes: 0GitHub stars: 9
- Llm Station SearchSearch code using LLM Station's indexed codebase. Faster and more accurate than grep for understanding code structure, finding symbols, and exploring dependencies.Votes: 0GitHub stars: 9
- Se Analyse StoryProduce a narrative shape analysis of a story — tension curve, plot points, pacing pattern, theme weights, character arc progress, and warnings. Use when the user asks "how's the pacing", "where's the tension sagging", "analyse this story", "does the midpoint hold".Votes: 0GitHub stars: 9
- Se Character TraceTrace a character through a story — their scenes, arcs, relationships, and where they appear. Use when the user asks "tell me about X", "what does character Y do", "show me Jonathan's arc", "where does Mina appear".Votes: 0GitHub stars: 9
- Se Check PlotInspect a plot thread — status, completion, phase, conflict, stakes, progress scoring — and list the scenes that advance it. Use when the user asks "how is the main plot going", "check on subplot X", "what's the state of the mystery thread".Votes: 0GitHub stars: 9
- Se Cli ReferenceComplete reference for the Story Engine CLI (./story-cli). Use when you need to know what CLI subcommand handles an operation or what flags a command accepts.Votes: 0GitHub stars: 9
- Se Edit SceneRevise an existing scene's prose or metadata. Use when the user asks "rewrite scene X", "fix the dialogue in Y", "update the summary for Z", or wants to replace narrativeText / dialogueText / actionText on an existing scene.Votes: 0GitHub stars: 9
- Se Enhance ProseAnalyse prose style (metrics + closest-register match) and optionally rewrite toward a target style or enhancement axis. Use when the user asks "is this scene Gideon-enough", "tighten this passage", "punch up the rhythm", "what register is this in", "rewrite scene X in Hemingway".Votes: 0GitHub stars: 9
- Se Explore StoryDeep dive into a single story — structure, characters, places, plots, arcs, and progress. Use when the user picks a story and says "tell me about X", "summarise", "what's in this story", or "give me the full picture".Votes: 0GitHub stars: 9
- Se List StoriesList every story in the Story Engine library with title, genre, word count, and id. Use when the user asks "what stories do I have", "show me my stories", "list stories", or needs a story id to feed into another skill.Votes: 0GitHub stars: 9
- Se Play StoryRead a story's scenes in order — narrative, dialogue, and action text. Use when the user asks "read me this story", "play through X", "walk me through the scenes", or wants to consume the prose end-to-end.Votes: 0GitHub stars: 9
- Se Progress ReportShow a story's progress dashboard — phase, completion %, open plot threads, arcs needing attention, word counts. Use when the user asks "how's the story going", "where am I in X", "what's the status".Votes: 0GitHub stars: 9
- Se Query TimelineBitemporal queries over a story — who was at a place, what happened when, a character's most recent action, a whole-story timeline. Use when the user asks "who was at X", "what did Y do last", "show me the timeline", "when did Z arrive at the inn".Votes: 0GitHub stars: 9
- Se Scene ContextPull every piece of already-established context for a single scene before writing prose. Use before generating or updating narrative/dialogue/action text so the prose stays consistent with the structure already in the database.Votes: 0GitHub stars: 9
- Se SearchFull-text search across every story's characters, scenes, places, and items. Use when the user asks "find", "where is", "search for", "look up", or names a character/place/scene without an id.Votes: 0GitHub stars: 9
- Se Write StoryCreate a new story end-to-end via the CLI — structure first, prose last. World → plot → cast → arcs → beats → prose, in that order. Use when the user asks "write me a story", "build a new story about X", "let's create a novella".Votes: 0GitHub stars: 9
- Start EngineStart the Story Engine. Handles port conflicts, Java detection via SDKMAN, and database setup automatically.Votes: 0GitHub stars: 9