Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/thomwebb/gac/adding-providernpx skills add thomwebb/gac --skill adding-providergit clone --depth 1 https://github.com/thomwebb/gacWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00078 | $0.02196 |
| Opus 5 | $0.00039 | $0.01098 |
| Sonnet 5 | $0.00016 | $0.00439 |
| Haiku 4.5 | $0.00008 | $0.00220 |
Grade A, and why
adding-provider scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 188 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Adding a Provider to GAC
Overview
GAC supports 35+ providers via a registry pattern. Adding one means creating a provider class, registering it, listing it in the interactive setup CLI, advertising it in the README, and adding standardized tests. Skipping any step leaves the provider partially wired (e.g. selectable in uvx gac init but not in the registry, or callable but not advertised).
Core principle: every provider lives in five places. Touch all five.
When to Use
Use when:
- A user asks to "add X as a provider"
- You need to support a new OpenAI-compatible inference service (cheapest case — extend
OpenAICompatibleProvider) - You need to support an Anthropic-compatible or custom-protocol provider
Do NOT use for:
- Modifying behavior of an existing provider (just edit its file)
- Adding a new model to an existing provider (no code change needed; users set
GAC_MODEL)
The Five Touch Points
| # | Path | Purpose |
|---|---|---|
| 1 | src/gac/providers/<key>.py |
Provider class (NEW file) |
| 2 | src/gac/providers/__init__.py |
Import + register_provider("<key>", Cls) |
| 3 | src/gac/model_cli.py |
Add to providers list in _configure_model; add alias mapping if display name → key derivation doesn't match |
| 4 | README.md |
Add to "Supported Providers" bullet list |
| 5 | tests/providers/test_<key>.py + tests/test_ai.py + tests/test_model_cli_providers.py |
Tests (NEW + edits) |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 188 lines · 78 tokens per session scan A 40537b6d5f07
adding-provider is a skill published in the GitHub repository thomwebb/gac (318 stars, last pushed 5d ago), licensed MIT. It adds 78 tokens to every session and 2,196 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
android-architecture
Default Android app architecture: MVVM, three Gradle modules, Jetpack Compose, Hilt, Coil, Ktor, Room, DataStore. Use when creating or changing Android, Kotlin, Jetpack Compose, Gradle modules, Activities, ViewModels, or Android libraries in an app that has not already chosen a different pattern.
seo-audit
SEO 诊断专家,基于 Google、Ahrefs、微软搜索指南设计的 92 项检查清单。 触发词:SEO审计、SEO诊断、网站SEO检查、为什么排名不好、技术SEO检查、页面SEO、E-E-A-T检查、内容质量分析。 输入一个网址,自动执行技术SEO(29项)、页面元素(27项)、内容质量与E-E-A-T(33项)、本地SEO(3项)四维度诊断,支持智能语言检测、站点分类与动态选页,生成中英文报告。.
running-the-agent
How to run miniagent locally, remotely, or in mixed mode, including Ollama, env vars, MCP, interactive REPL, and provider fallback. Use when running agent.py, adding a provider, debugging tool calling, or setting up .env.
ios-architecture
Default iOS and macOS app architecture: MVVM, SwiftUI, @Observable, async/await, URLSession. Use when creating or changing iOS, iPadOS, macOS, SwiftUI, ViewModels, Swift packages, or Xcode app targets that have not already chosen a different pattern.
new-project
Bootstrap a new product repository for agentic development. Use when starting a new app, empty repo, greenfield project, or when the user asks to stand up the factory, the start loop, or what to install before the first feature.
web-architecture
Default web app architecture: TypeScript, feature folders, a view / state / client split in the same spirit as MVVM. Use when creating or changing web UI, TypeScript/JavaScript frontend, React or similar SPA/SSR pages, CSS, or browser API clients in an app that has not already chosen a different pattern.