提示词
提示词工程与模板
avp
portofcontext
Use this skill for ANY work on AVP (Agent Voyager Project): agents, supervisor Commissions, conformance cases, or debugging AVP runs. AVP wires agents (drive an LLM, emit events) to supervisors (declare environment).
pydantic-ai-agents
Fuenfgeld
"Build and debug Pydantic AI agents using best practices for dependencies, dynamic system prompts, tools, and structured output validation. Use when the user wants to: (1) Create a new Pydantic AI agent, (2) Debug or fix an existing agent, (3) Add features like tools, validators, or dynamic prompts, (4) Integrate OpenRouter for multi-model access, (5) Add Logfire for debugging/observability, (6) Structure agent architecture with dependency injection."
pydantic-evals
Fuenfgeld
"Test and evaluate AI agents and LLM outputs using code-first evaluation framework with strong typing. Use when the user wants to: (1) Create evaluation datasets with test cases for AI agents, (2) Define evaluators (deterministic, LLM-as-Judge, custom, or span-based), (3) Run evaluations and generate reports, (4) Compare model performance across experiments, (5) Integrate evaluations with Pydantic AI agents, (6) Set up observability with Logfire, (7) Generate test datasets using LLMs, (8) Implement regression testing for AI systems."
agentproto-llm-endpoint
agentproto
Use the local LLM endpoint proxy (localhost:18090) — or your own gated public
pb-supervise-parallel-mission
agentproto
Run N parallel worker agents on disjoint file/topic ownership and fan their results into one verified deliverable. Trigger when supervising - 'I am the supervisor', 'spawn parallel workers', 'fan-in several agents', 'decompose this mission and delegate'.
ap-prompt-agent
agentproto
Send a follow-up prompt to a live agentproto session, interrupt its in-flight turn, switch its model/effort/posture without restarting, or manage its queued prompts. Use when the user says "tell the agent to also...", "interrupt and redirect it", "switch that session to a cheaper model", "promote/skip that queued prompt", or "cancel that queued message".
pb-nested-orchestrator
agentproto
Spawn a parent agent that spawns and supervises its own sub-agents, forming a two-level tree on the agentproto daemon. Trigger for delegation depth - 'agent that spawns sub-agents', 'two-level agent tree', 'let an agent orchestrate on its own'.
instruction-engineering
axiomantic
"Use when crafting, improving, or reviewing prompts, system prompts, skill instructions, or any text that instructs an LLM. Triggers: 'write a prompt', 'prompt engineering', 'improve this prompt', 'design a system prompt', 'write skill instructions', 'craft agent instructions'. Provides CSO (Claude Search Optimization) guidance for skill descriptions. Also invoked by writing-skills."
reinforce-skills
b-open-io
"This skill should be used when the user asks to 'reinforce skills', 'add skill map', 'update skill map', 'sync skills to CLAUDE.md', 'persist skills', 'save skills to project', 'embed skills', 'skills keep getting forgotten', 'I keep forgetting skills', or when setting up a new project where installed skills should be persisted as context. Generates a compressed skill-mapping directive in CLAUDE.md following the Vercel AGENTS.md research pattern."
table-image
maroffo
"Render markdown tables as hand-drawn sketch images. Use when user wants a table rendered as an image, visual table, or diagram illustration."
hook-manager
b-open-io
Discover and install automation hooks for Claude Code and Opencode. This skill should be used when users ask to "list hooks", "install a hook", "show available hooks", "enable hook", "what hooks are available", or need help managing agent automation hooks.
obliteratus
Russ534
"OBLITERATUS: abliterate LLM refusals (diff-in-means)."
gemini-review
maroffo
"Local code review with Gemini CLI for pre-commit reviews. Use when user says review code, code review, check my changes, or review before commit."
agent-eval
SeuPut0705
Head-to-head comparison of coding agents (Claude Code, Aider, Codex, etc.) on custom tasks with pass rate, cost, time, and consistency metrics. Use when choosing between coding agents, or when a change to an agent setup needs measured pass rate, cost, and time rather than an impression.
housekeeping
bout3fiddy
Repository housekeeping workflows for AGENTS/CLAUDE architecture, progressive disclosure, and migration of legacy monolithic instruction files.
agent-architecture-audit
SeuPut0705
Full-stack diagnostic for agent and LLM applications. Audits the 12-layer agent stack for wrapper regression, memory pollution, tool discipline failures, hidden repair loops, and rendering corruption. Produces severity-ranked findings with code-first fixes. Essential for developers building agent applications, autonomous loops, or any LLM-powered feature. Use when an agent or LLM feature misbehaves and the failing layer is unknown, or before shipping an agent stack.
inbox-triage
maroffo
Review and prioritize Gmail inbox. Use when user wants to check email, review inbox, or see what needs attention.
manage-agent
t-benze
Enroll, update, or terminate an agent. Write a JSON file and call happyranch manage-agent --from-file to keep the invocation single-line. Enrollment requires founder approval.
cost-aware-llm-pipeline
SeuPut0705
Cost optimization patterns for LLM API usage — model routing by task complexity, budget tracking, retry logic, and prompt caching. Use when LLM spend needs to come down, or when routing tasks across model tiers and budgets.
skill-creator
XiaoMi
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
foundation-models-on-device
SeuPut0705
Apple FoundationModels framework for on-device LLM — text generation, guided generation with @Generable, tool calling, and snapshot streaming in iOS 26+. Use when adding on-device LLM features with Apple FoundationModels on iOS 26+.
neuro-symbolic-reasoning
sundial-org
Neuro-symbolic AI combining LLMs with symbolic solvers. Use when exploring neuro-symbolic approaches (ideation, no code) or implementing solver integrations (code).
tinker-training-cost
sundial-org
Calculate training costs for Tinker fine-tuning jobs. Use when estimating costs for Tinker LLM training, counting tokens in datasets, or comparing Tinker model training prices. Tokenizes datasets using the correct model tokenizer and provides accurate cost estimates.
obsidian-clipper-template-creator
GuppyTheCat
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.