Aggregate pre-computed metrics (gauge, counter, delta types) using OPAL. Use when analyzing request counts, error rates, resource utilization, or any numeric metrics over time. Covers align + m() + aggregate pattern, summary vs time-series output, and common aggregation functions. For percentile metrics (tdigest), see analyzing-tdigest-metrics skill.
Work with Resource datasets (mutable state tracking) using OPAL temporal joins. Use when you need to enrich Events/Intervals with contextual state information, track resource state changes over time, or navigate between datasets using temporal relationships. Covers temporal join mechanics (lookup, join, follow), automatic field matching, and when to use Resources vs Reference Tables.
Use OPAL subquery syntax (@labels) and union operations to combine multiple datasets or time periods. Essential for period-over-period comparisons, multi-dataset analysis, and complex data transformations. Covers @label <- @ syntax, timeshift for temporal shifts, union for combining results, and any_not_null() for collapsing grouped data.
Work with Reference Tables (static CSV lookup data) using OPAL to enrich datasets with descriptive information. Use when you need to map IDs to human-readable names, add static metadata from CSV uploads, or perform lookups without temporal considerations. Covers both explicit and implicit lookup patterns, column name matching, and when to choose Reference Tables vs Resources vs Correlation Tags.
Analyze event datasets (logs) and intervals over time using OPAL timechart. Use when you need to visualize trends, track metrics over time, or create time-series charts. Covers timechart for temporal binning, bin duration options (1h, 5m, 1d), options(bins:N) for controlling bin count, and understanding temporal output columns (_c_valid_from, _c_valid_to, _c_bucket). Returns multiple rows per group for time-series visualization. For single summaries, see aggregating-event-datasets skill.
Master OPAL window functions for row-relative calculations, rankings, and moving aggregates. Covers lag(), lead(), row_number(), rank(), dense_rank(), moving averages, first(), and last(). Use when comparing rows to neighbors, ranking within partitions, calculating rate of change, or computing time-based moving windows. CRITICAL - OPAL uses window() function wrapper, NOT SQL OVER clause.
Stream structured content to persistent SQLite storage with automatic session break recovery. Core principle: The content IS the state. Every stream_write is automatically persistent. Supports multi-format export (Markdown, HTML, JSON, YAML, CSV, Text) and 7 document templates. Commands: /stream-init, /stream-status, /stream-read, /stream-write, /stream-export. ALWAYS call stream_status after session breaks to check for resume_from and preserved_context.
Extract structured fields from unstructured log data using OPAL parsing functions. Covers extract_regex() for pattern matching with type casting, split() for delimited data, parse_json() for JSON logs, and JSONPath for navigating parsed structures. Use when you need to convert raw log text into queryable fields for analysis, filtering, or aggregation.
Create and manage datasets on Hugging Face Hub. Supports initializing repos, defining configs/system prompts, and streaming row updates. Designed to work alongside HF MCP server for comprehensive dataset workflows.
Manage PostgreSQL databases using Postgres MCP. Query data, inspect schemas, analyze table structures, run migrations, debug database issues, and manage test data. Use when working with databases, debugging queries, or validating data integrity.
Processes CSV files and pandas DataFrames. Use when working with CSV files, tabular data, spreadsheets, or when the user asks to query, analyze, or manipulate structured data.
Filter and search event datasets (logs) using OPAL. Use when you need to find specific log events by text search, regex patterns, or field values. Covers contains(), tilda operator ~, field comparisons, boolean logic, and limit for sampling results. Does NOT cover aggregation (see aggregating-event-datasets skill).
Design effective tools for AI agents - tool descriptions, consolidation principles, architectural reduction, MCP naming conventions, and error handling
Export and deploy fine-tuned models to production. Covers GGUF/Ollama, vLLM, HuggingFace Hub, Docker, quantization, and platform selection. Use after fine-tuning when you need to deploy models efficiently.
Create, validate, install, convert, port, and manage Claude Code skills. Use when users want to create a new skill, validate existing skills, install from GitHub, list skills, convert MCP servers to skills, or port skills between Claude and Gemini platforms. Covers the full skill lifecycle from creation to cross-platform distribution.
Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).
Coordinates Serena MCP knowledge graph operations for Shannon Framework. Enforces standardized entity naming (shannon/* namespace), relation creation patterns, search protocols, and observation management. Prevents orphaned entities, naming chaos, and broken context lineage. Use when: storing specs/waves/goals/checkpoints, querying Shannon history, managing knowledge graph structure, ensuring cross-wave context preservation.
Comprehensive guide for building production-ready agents with the Claude Agent SDK. Use when creating agents, designing tools, implementing subagents, managing sessions, integrating MCP servers, or understanding SDK-native features. Emphasizes documentation-first approach and using only SDK native capabilities.
MCP server development patterns extending Anthropic's mcp-builder with AINative-specific conventions. Use when creating MCP servers, integrating ZeroDB, or building tool-based AI systems.
Project-specific expertise for the tools-automation monorepo. Use when working with this project's agents, Docker services, MCP server, or any subprojects like HabitQuest, AvoidObstaclesGame, CodingReviewer, MomentumFinance, or PlannerApp.
