A skill that runs a hard, testable coding task through a multi-agent harness instead of a single model pass (base variant, no external model needed). It takes implementations, large refactors, or migrations whose correctness can be expressed as tests and runs them through plan → parallel variants (with a test-driven self-correction loop) → adversarial verification (Claude itself) → synthesis, raising correctness over a single pass. Works on Claude Code's Workflow orchestration alone. Use when the user says "pantheon", "run it through pantheon", "build several variants, gate on tests, then adversarially verify", "really hammer this implementation", or wants to solve a hard coding/agentic task with multiple agents. For cross-model (GPT-5.5) adversarial verification, use the pantheon-x skill instead. Don't use for easy one-shot questions or trivial edits (cost is high).
EnglishA skill that runs a hard, testable coding task through a multi-agent harness (configurable variant: the user PICKS which AI model runs the adversarial verification — including non-Anthropic models like DeepSeek, Qwen, Kimi, or a local Ollama/LM Studio model). Same plan → parallel variants (test-driven self-correction) → adversarial verification → synthesis pipeline as the pantheon base, but the verifier is chosen per run via a verifier argument and external/local models are driven through the codex CLI (which is itself a multi-provider router). Use when the user says "pantheon custom", "verify with deepseek/qwen", "use DeepSeek as the grader", "have a local model break it", "외부 모델로 검증", "딥시크로 채점", "큐원으로 검증", "오픈클로처럼 채점자 모델 골라서", "로컬 모델로 검증". If no model is given it defaults to Claude (same as the pantheon base). For the fixed presets use pantheon (Claude) / pantheon-x (GPT-5.5). For REVIEWING an existing project rather than generating code, use pantheon-gap-custom. Don't use for easy one-shot work (cost is high).
EnglishA skill that runs a GAP ANALYSIS & feedback review of an existing project through a multi-agent harness instead of a single review pass.
EnglishA skill that runs a GAP ANALYSIS & feedback review of an existing project through a multi-agent harness (configurable variant: the user PICKS which AI model runs the…
EnglishA skill that runs a GAP ANALYSIS & feedback review of an existing project through a multi-agent harness (enhanced variant: GPT-5.5 cross-model adversarial…
EnglishConfigures WHICH AI model runs the adversarial step in pantheon-custom / pantheon-gap-custom — the OpenClaw-style "pick your model" setup, as its own command. It lists the models actually available on this machine (Claude tiers, GPT-5.5 via Codex, local Ollama/LM Studio models, and the cloud catalog in providers.json — DeepSeek, Qwen, Gemini, Mistral, Groq, xAI/Grok, Kimi, ...), lets the user pick one (OpenClaw-style provider/model-id), securely sets up the API key in a file (never in chat) when a cloud model needs one, and saves the choice to ~/.pantheon/config.json so the custom skills use it without asking again. Use when the user says "pantheon model", "pantheon-model", "pick/choose the pantheon verifier model", "change the pantheon model", "configure pantheon custom", "팬테온 모델 설정", "모델 고르기/다시 고를래", "채점 모델 바꿔". This skill only CONFIGURES; to actually run a job use pantheon-custom (generate) or pantheon-gap-custom (review).
EnglishA skill that runs a hard, testable coding task through a multi-agent harness (enhanced variant: GPT-5.5 cross-model adversarial verification). It takes implementations, large refactors, or migrations whose correctness can be expressed as tests and runs them through plan → parallel variants (test-driven self-correction loop) → GPT-5.5 (Codex) adversarial verification → synthesis. Because the implementation Claude wrote is attacked by a different model, it's stricter than the base (pantheon). Requires a Codex CLI login. Use when the user says "pantheon x", "pantheon enhanced", "GPT-5.5 adversarial verify", "cross-model hard verification", or wants the strongest cross-verified pipeline. If Codex/GPT-5.5 isn't available, use the pantheon skill instead. Don't use for easy one-shot work (cost is high).
English
AGENT SKILLS ATLASIncludes Pantheon Custom, Pantheon Gap, Pantheon Gap Custom, and other workflows.
Claude Code
Codex
Cursor
GitHub Copilot
Gemini CLISupported by the Atlas install prompt. Skill-specific requirements vary.
Pack stats
What’s included
Pantheon
A skill that runs a hard, testable coding task through a multi-agent harness instead of a single model pass (base variant, no external model needed). It takes implementations, large refactors, or migrations whose correctness can be expressed as tests and runs them through plan → parallel variants (with a test-driven self-correction loop) → adversarial verification (Claude itself) → synthesis, raising correctness over a single pass. Works on Claude Code's Workflow orchestration alone. Use when the user says "pantheon", "run it through pantheon", "build several variants, gate on tests, then adversarially verify", "really hammer this implementation", or wants to solve a hard coding/agentic task with multiple agents. For cross-model (GPT-5.5) adversarial verification, use the pantheon-x skill instead. Don't use for easy one-shot questions or trivial edits (cost is high).
Pantheon Custom
A skill that runs a hard, testable coding task through a multi-agent harness (configurable variant: the user PICKS which AI model runs the adversarial verification — including non-Anthropic models like DeepSeek, Qwen, Kimi, or a local Ollama/LM Studio model). Same plan → parallel variants (test-driven self-correction) → adversarial verification → synthesis pipeline as the pantheon base, but the verifier is chosen per run via a verifier argument and external/local models are driven through the codex CLI (which is itself a multi-provider router). Use when the user says "pantheon custom", "verify with deepseek/qwen", "use DeepSeek as the grader", "have a local model break it", "외부 모델로 검증", "딥시크로 채점", "큐원으로 검증", "오픈클로처럼 채점자 모델 골라서", "로컬 모델로 검증". If no model is given it defaults to Claude (same as the pantheon base). For the fixed presets use pantheon (Claude) / pantheon-x (GPT-5.5). For REVIEWING an existing project rather than generating code, use pantheon-gap-custom. Don't use for easy one-shot work (cost is high).
Pantheon Gap
A skill that runs a GAP ANALYSIS & feedback review of an existing project through a multi-agent harness instead of a single review pass. It maps the project, fans out one
Pantheon Gap Custom
A skill that runs a GAP ANALYSIS & feedback review of an existing project through a multi-agent harness (configurable variant: the user PICKS which AI model runs the adve
Pantheon Gap X
A skill that runs a GAP ANALYSIS & feedback review of an existing project through a multi-agent harness (enhanced variant: GPT-5.5 cross-model adversarial confirmation)
Pantheon Model
Configures WHICH AI model runs the adversarial step in pantheon-custom / pantheon-gap-custom — the OpenClaw-style "pick your model" setup, as its own command. It lists the models actually available on this machine (Claude tiers, GPT-5.5 via Codex, local Ollama/LM Studio models, and the cloud catalog in providers.json — DeepSeek, Qwen, Gemini, Mistral, Groq, xAI/Grok, Kimi, ...), lets the user pick one (OpenClaw-style provider/model-id), securely sets up the API key in a file (never in chat) when a cloud model needs one, and saves the choice to ~/.pantheon/config.json so the custom skills use it without asking again. Use when the user says "pantheon model", "pantheon-model", "pick/choose the pantheon verifier model", "change the pantheon model", "configure pantheon custom", "팬테온 모델 설정", "모델 고르기/다시 고를래", "채점 모델 바꿔". This skill only CONFIGURES; to actually run a job use pantheon-custom (generate) or pantheon-gap-custom (review).
Pantheon X
A skill that runs a hard, testable coding task through a multi-agent harness (enhanced variant: GPT-5.5 cross-model adversarial verification). It takes implementations, large refactors, or migrations whose correctness can be expressed as tests and runs them through plan → parallel variants (test-driven self-correction loop) → GPT-5.5 (Codex) adversarial verification → synthesis. Because the implementation Claude wrote is attacked by a different model, it's stricter than the base (pantheon). Requires a Codex CLI login. Use when the user says "pantheon x", "pantheon enhanced", "GPT-5.5 adversarial verify", "cross-model hard verification", or wants the strongest cross-verified pipeline. If Codex/GPT-5.5 isn't available, use the pantheon skill instead. Don't use for easy one-shot work (cost is high).
Quick start
- Choose your skills. Select individual skills above, or choose the entire pack.
- Copy the install prompt. Use Copy install prompt above the list.
- Send it to your agent. Paste into your coding agent’s chat. Your agent reviews the source and guides the setup.
Try it with your agent
After installation, start with a request like:
Use the Pantheon skill to help with my task.
Ask me for the context you need before getting started.Requirements & license
- A coding agent that can access your project files and install skills.
- Your agent checks each skill’s instructions, dependencies, and permissions during setup.
- Reported license for the skills shown: MIT. The author’s terms apply.