Loading...
Loading...
Found 43 Skills
Reviewing one Task's diff on two axes in parallel sub-agents — Spec (does the code do what the spec and the ticket asked?) and Standards (does it follow the repo's conventions, the test rules and the smell baseline?) — and reporting them side by side. The reviewer step of /archie-implement.
Illustre automatiquement le journal d'une aventure BFRPG en générant des images pour les moments clés (combats, explorations, découvertes). Utilise la génération parallèle pour une performance optimale.
This skill should be used at the start of any computationally intensive scientific task to detect and report available system resources (CPU cores, GPUs, memory, disk space). It creates a JSON file with resource information and strategic recommendations that inform computational approach decisions such as whether to use parallel processing (joblib, multiprocessing), out-of-core computing (Dask, Zarr), GPU acceleration (PyTorch, JAX), or memory-efficient strategies. Use this skill before running analyses, training models, processing large datasets, or any task where resource constraints matter.
Novel Logic/Plot Review, applicable to user requests such as "Help me check if there are bugs in my novel", "Check if there are timeline contradictions", "Check if characters are OOC", "Find plot conflicts between different parts", "Sort out whether foreshadowings are resolved", "Check the rationality of novel plots", "Check if there are plot loopholes", "Character behaviors are inconsistent with their personalities", "Check if the timeline is correct", "Find contradictions in the novel", "Help me sort out all foreshadowings", "Novel plot bug check", "Logical loophole troubleshooting", etc. It detects issues such as timeline conflicts, logical loopholes, character OOC, and missing foreshadowings. **Performs word count checks to ensure chapter word counts meet standards**, **generates a detailed issue list and automatically fixes all issues, with the fixed results directly modifying the chapters/ directory and automatically backing up the original files to .sumeru/write/original/ before modification**, **uses sub-Agents for parallel processing during batch review, with each Agent responsible for a maximum of 3 chapters**
AI agent patterns with Trigger.dev - orchestration, parallelization, routing, evaluator-optimizer, and human-in-the-loop. Use when building LLM-powered tasks that need parallel workers, approval gates, tool calling, or multi-step agent workflows.
Analyze OpenCode conversation history to identify themes and patterns in user messages. Use when asked to analyze conversations, find themes, review how a user steers agents, or extract insights from session history.
Fan out a prompt to multiple AI coding agents in parallel and synthesize their responses.
Deep Performance Optimization Skill for Triton Operators on Ascend NPU, dedicated to achieving the Triton operator performance improvement required by users. Core technologies include but are not limited to Unified Buffer (UB) capacity planning, multi-Tokens parallel processing, MTE/Vector pipeline parallelism, mask optimization, etc. This Skill must be triggered when the user mentions the following: performance optimization of Vector-type Triton operators on Ascend NPU.
Use when batch-resolving approved todos, especially after code review or triage sessions
Write node content documents. Read download.txt, integrate local materials for each node and write detailed, accurate, and complete Markdown documents. Each sub-agent processes one node in parallel, outputting a complete node document including overview, directory/mind map, flow chart, online image URL, and reference materials. Suitable for scenarios requiring systematic and structured content creation.
Worker that runs parallel external agent reviews (Codex + Gemini) on Story/Tasks. Background tasks, process-as-arrive, critical verification with debate. Returns filtered suggestions for Story validation.
Multi-instance (Multi-Agent) orchestration workflow for deep research: Split a research goal into parallel sub-goals, run child processes in the default `workspace-write` sandbox using Codex CLI (`codex exec`); prioritize installed skills for networking and data collection, followed by MCP tools; aggregate sub-results with scripts and refine them chapter by chapter, and finally deliver "finished report file path + key conclusions/recommendations summary". Applicable to: systematic web/data research, competitor/industry analysis, batch link/dataset shard retrieval, long-form writing and evidence integration, or scenarios where users mention "deep research/Deep Research/Wide Research/multi-Agent parallel research/multi-process research".