# Agent Toolbox Reusable agent skills for project setup, behavior improvement, and safe repository maintenance. This repository root is a self-contained [Agent Plugins 1.0](https://agent-plugins.org/specification) package. Compatible clients load the portable `plugin.json` manifest and discover each immediate child under `skills/` as an Agent Skill. The package is skills-first on purpose: detailed workflow instructions load only when a task matches a focused skill. ## Install Follow the active client's Agent Plugins setup instructions and select this repository root as the plugin directory. Distribution, installation, enablement, and updates are client-specific rather than part of the portable specification. A client may need a separate marketplace entry or manifest for its installation flow; add such an adapter only for a documented target requirement while keeping this portable package as the source of the skills. See [docs/setup.md](docs/setup.md) for package use and project-instruction guidance. See [tests/behavior/README.md](tests/behavior/README.md) for regression checks and agent evaluation. ## Layout | Path | Purpose | | --- | --- | | `plugin.json` | Portable Agent Plugins 1.0 manifest. | | `skills/` | Agent Skill entrypoints bundled with this plugin. | | `memory/` | Reusable memory templates and schema, not project checkpoints. | | `prompts/` | Reusable prompt templates. | | `mcp/` | Token-free MCP examples and config snippets; not a runtime `mcp.json` component. | | `docs/` | Setup and operating guides. | ## Bundled Skills | Skill | Purpose | | --- | --- | | `audit-agent-assets` | Audit and modernize agent-facing assets for context efficiency and current model capabilities. | | `prevent-repeat` | Investigate missed behavior and apply durable prevention. | | `rename-master-to-main` | Migrate `master` to `main` with verified history preservation and guarded cleanup. | | `setup-agent-project` | Discover client guidance and refresh evidence-driven project instructions idempotently. | ## How To Use After installing the plugin, start a new task or chat and invoke the skill that matches the work. ```text Use $audit-agent-assets to review this repository's agent-facing assets and recommend changes only when the evidence supports them. ``` ```text Use $prevent-repeat to investigate why this instruction was missed and recommend a durable fix. ``` ```text Use $setup-agent-project to configure this repository for agent-assisted work. ``` Keep secrets, credentials, local OAuth state, and project-specific checkpoints out of this repository. Project memory belongs in the project using the plugin.