MadPonyInteractive

MPI workflow pack - shared reference library for the mpi-kanban skills. Agents may read its reference files directly for card/task-board rules; do not run it as a user workflow.

MPI workflow pack - Continue active MPI work, show/read one board task, or update one task-card state. Use when the user says "continue this MPI plan", "MPI continue", "continue", "resume", "keep going", "pick this back up", "read a handoff and continue",…

MPI workflow pack - Execute a parallel batch with worker sub-agents, from a plan's `## Parallel Batch` section or from ready cards on the JSON task board. Use when a plan contains a `## Parallel Batch` section with task ownership, or when the user says "MPI…

MPI workflow pack - Extract and return Sub-Agent Briefing text from configured project rules or rule bundles. Use when dispatching sub-agents, resolving worker briefings, or when the user invokes "mpi-brief-rule <name>".

MPI workflow pack - Maintain an existing Mpi-Kanban project. Audits project knowledge, board and state consistency, rules, memory pointers, important commands, architecture summary, source-of-truth mode, and legacy migration drift. Use when the user says "MPI…

MPI workflow pack - Close out finished MPI work. Syncs rules and docs, heals project knowledge, writes memory, resolves every validating card, commits and pushes, and closes the JSON task card. Use when the user says "end session", "wrap up", "commit and…

MPI workflow pack - Hand the work to a fresh session in under two minutes. Commits and pushes the session's files, writes a handoff JSON from the running notes already in the active plan, and prints a paste-ready resume block. Use when the user says…

MPI workflow pack - Review the task board and fold related cards into an umbrella card. Use when the user says "create an umbrella", "make an umbrella card", "umbrella these cards", "evaluate the cards on the kanban", "review the board and group the cards",…

Make the two preview assets a Flow ships — the 4/5 tile still and the wide autoplaying hero clip. Use when the user says "flow graphics", "create graphics for <flow>", "let's do the graphics for <flow>", "make the images for the <name> flow", "the flow needs…

Work on the first-party ComfyUI node pack (ComfyUi-MpiNodes) and keep Cubric Vision's pin in sync. Use when the user says "add a node", "new comfy node", "update a node", "change MpiNodes", "the nodes repo changed", "pin the nodes", "MpiNodes drift", "sync…

Bump the ComfyUI engine that SHIPS TO USERS (dev_configs/node_lock.json comfyui.core.tag + system_dependencies.json), and smoke-test models by executing a minimal generation per op on a RunPod Pod. Use when the user says "bump ComfyUI", "bump the engine", "we…

Cut a Cubric Vision release — the single GitHub-only release flow. Bump the right version digit (2nd = new features/ops/engine, 3rd = bug fixes, 1st = breaking), stamp the files via mpi-version-bump, build the portable artifacts in CI, and publish a GitHub…

The file-edit MECHANIC for a Cubric Vision version bump — bump app version (appVersion.js + package.json + package-lock.json), update operation registry, update model mappings, sync operation_registry.json, generate release notes, run release:check, and offer…

Wire a NEW Flow (dev-gated Flow-Library outcome flow) into Cubric Vision end-to-end — descriptor + universal op in 4 files + workflow + optional uiComponent + media I/O + reuse. Use when the user says "add a new flow", "let's add a flow", "wire up the <name>…

Wire a NEW model into Cubric Vision end-to-end — workflow template→runtime split, dependencies.js entries, R2 upload, models.js ModelDef, progressStages, type-consumer sweep, style-LoRA system. Use when the user says "add a new model", "let's add <model>",…

Upgrade the standalone LOCAL ComfyUI authoring bench (G:\ComfyUi) to a target version and keep its hand-maintained extra_model_paths.yaml in sync with the app's dep folder types. Use when the user says "bump my local ComfyUI", "update my test ComfyUI",…