Loading...
Loading...
Found 43 Skills
Generate a self-contained HTML viewer for any Claude Code session, including agent team sessions with full inter-agent DM timelines. Use whenever the user asks to "view a session", "visualize a conversation", "show me what happened in session X", "generate a session viewer", "replay a session", or references viewing/inspecting Claude Code JSONL logs. Also use when the user provides a session ID and wants to see the conversation.
Core patterns for AI coding agents based on analysis of Claude Code, Codex, Cline, Aider, OpenCode. Triggers when: Building an AI coding agent or assistant, implementing tool-calling loops, managing context windows for LLMs, setting up agent memory or skill systems, or designing multi-provider LLM abstraction. Capabilities: Core agent loop with while(true) and tool execution, context management with pruning and compression and repo maps, tool safety with sandboxing and approval flows and doom loop detection, multi-provider abstraction with unified API for different LLMs, memory systems with project rules and auto-memory and skill loading, session persistence with SQLite vs JSONL patterns.
Use when diagnosing unexpected behavior, failed workflows, bugs, browser or Node.js runtime issues, logs, traces, or when preparing a root-cause hypothesis. 诊断异常、定位 bug、判断修复方向时使用：先建立证据表，区分运行时事实和代码推断，避免多层猜测；证据不足时添加 copy-friendly 浏览器日志或本地 Node.js JSONL 日志。
Use when debugging a Nemo Gym run or reward profiling job. Covers rollout collection failures, empty or partial JSONL outputs, stale materialized inputs, verifier/schema errors, Ray or Slurm issues, vLLM readiness, judge failures, tool/sandbox failures, cache problems, and throughput bottlenecks.
Auto-capture per-session token usage from the Claude Code session jsonl and persist to the cost-tracking namespace
Synthesizes trust boundaries, attack surfaces, and attacker profiles into a living threat model. Use as Stage B of the Knowledge Base generation process, reading architecture and entity definitions from the KB. Don't use for analyzing source code or extracting raw learnings from JSONL files.
Local token cost analytics dashboard for Claude Code sessions — reads JSONL transcripts and provides per-prompt cost breakdowns, heatmaps, and usage insights.
Building & extending Pi — authoring TypeScript extensions (ExtensionAPI, registerTool, registerProvider, /commands, UI hooks), publishing as npm/git packages (pi-package), embedding via JSON-RPC mode (--mode rpc/json, JSONL framing, AgentSession SDK), and developing inside the pi_agent_rust repo. Use for any "how do I build a Pi extension/package/SDK client" question.
Turn a brief music description and optional tagged lyrics into a professional MiniMax Music 3 structured caption with Global Metadata, Vocal Details, and a section-aware Arrangement. Use when users ask to enhance a music-generation prompt, preserve lyric-section directives, retrieve a similar style from bundled templates, fuse styles, or produce JSON or JSONL caption output.
Use to help users get started with Nemo Gym reward profiling. Covers the basic ng_run, ng_collect_rollouts, and ng_reward_profile workflow, repeated rollouts, materialized inputs, rollout JSONL artifacts, task and rollout identity, output inspection, partial profiling, and rollout_infos. For failed jobs, prefer nemo-gym-debugging.
Manage local Codex session transcripts, including listing candidate sessions, exporting full or selected sessions to organized Markdown, inspecting archived sessions, and summarizing tool-call history. Use when the user asks to scan, parse, archive, inspect, recover, summarize, manage, or convert Codex sessions, `~/.codex/sessions` data, `~/.codex/archived_sessions` data, `.jsonl` transcripts, tool-call history, or hard-to-read Codex conversation logs.
Archive deployment records with merged common+environment config context (including remote port) for Makefile-first deployment workflow.