From aaf47fb4e0b3e5814892518805daa4ff2f4b09cb Mon Sep 17 00:00:00 2001 From: HiranoMasaaki Date: Thu, 12 Feb 2026 18:27:51 +0900 Subject: [PATCH 1/2] Docs: Replace "declarative TOML" with natural-language phrasing The file format is an implementation detail. Reword to emphasize that Experts are defined in natural language, not in a specific format. Co-Authored-By: Claude Opus 4.6 --- README.md | 4 ++-- docs/README.md | 2 +- docs/contributing/roadmap.md | 2 +- docs/index.md | 8 ++++---- 4 files changed, 8 insertions(+), 8 deletions(-) diff --git a/README.md b/README.md index 4e223960..427e5474 100644 --- a/README.md +++ b/README.md @@ -1,4 +1,4 @@ -# Perstack: Agentic AI Runtime +# Perstack: The Agentic AI Runtime

npm version @@ -16,7 +16,7 @@ - **`perstack` CLI**: Develop, test, and deploy Experts from the command line. - **`create-expert` app**: Creates, tests, and refines new Experts in `perstack.toml` to your specification — powered by agentic AI. -- **`perstack.toml`**: Define and orchestrate reusable, domain-specialized agents as Experts in declarative TOML. +- **`perstack.toml`**: Define and orchestrate reusable, domain-specialized agents as Experts in natural language. - **`@perstack/runtime`**: Execute Experts programmatically. Event-sourced state with step-level checkpoints — resume, replay, and audit any execution from any point. ## Quick Start diff --git a/docs/README.md b/docs/README.md index 8d3452b7..306a1cd5 100644 --- a/docs/README.md +++ b/docs/README.md @@ -2,7 +2,7 @@ title: "Perstack: The Agent Runtime" --- -Define AI agents as declarative **Experts** in TOML. Execute them with deterministic, event-derived tracking. Each Expert runs in its own isolated context — no shared state, no prompt bloat, full execution history. +Define AI agents as **Experts** in natural language. Execute them with deterministic, event-derived tracking. Each Expert runs in its own isolated context — no shared state, no prompt bloat, full execution history. - [Getting Started →](./getting-started/walkthrough.md) - [Browse Registry →](https://platform.perstack.ai/) diff --git a/docs/contributing/roadmap.md b/docs/contributing/roadmap.md index 73810f51..6859cc29 100644 --- a/docs/contributing/roadmap.md +++ b/docs/contributing/roadmap.md @@ -9,7 +9,7 @@ This roadmap outlines the direction of Perstack development. Priorities may shif | Feature | Status | Version | Timeline | Description | | ------------------------------ | :----: | ------- | --------- | ------------------------------------------------- | -| **Expert definitions** | ✅ | v0.0 | - | Declarative definitions in `perstack.toml` | +| **Expert definitions** | ✅ | v0.0 | - | Natural-language definitions in `perstack.toml` | | **Runtime** | ✅ | v0.0 | - | Agent loop, checkpoint/restore, event sourcing | | **Registry** | ✅ | v0.0 | - | Public registry with write-once versioning | | **Skills (MCP)** | ✅ | v0.0 | - | MCP-based tool integration | diff --git a/docs/index.md b/docs/index.md index 840d0937..9dfdc55b 100644 --- a/docs/index.md +++ b/docs/index.md @@ -1,11 +1,11 @@ --- title: Perstack Documentation -description: Define AI agents as declarative Experts in TOML. Execute them with deterministic, event-derived tracking. +description: Define AI agents as Experts in natural language. Execute them with deterministic, event-derived tracking. slug: docs pagefind: true --- -Define AI agents as declarative **Experts** in TOML. Execute them with deterministic, event-derived tracking. Each Expert runs in its own isolated context — no shared state, no prompt bloat, full execution history. +Define AI agents as **Experts** in natural language. Execute them with deterministic, event-derived tracking. Each Expert runs in its own isolated context — no shared state, no prompt bloat, full execution history. ## Quick Start @@ -21,11 +21,11 @@ npx perstack start code-reviewer "Review PR #42 for security issues" ## Why Perstack? -Agent frameworks help you **build** agents. Perstack is a **runtime** that **executes** them. You define Experts declaratively in TOML — Perstack handles execution, isolation, and state. +Agent frameworks help you **build** agents. Perstack is a **runtime** that **executes** them. You define Experts in natural language via TOML — Perstack handles execution, isolation, and state. | | | | :--- | :--- | -| **Declarative definitions** | Expert definitions in `perstack.toml` are plain text. Domain experts write the behavior; developers focus on integration. | +| **Natural-language definitions** | Expert definitions in `perstack.toml` are plain text. Domain experts write the behavior; developers focus on integration. | | **Isolation by design** | Each Expert runs in its own context — workspace boundaries, environment sandboxing, and tool whitelisting. No shared state between runs. | | **Event-derived execution** | Step-level checkpoints make runs reproducible. Resume from any step, debug, replay, and audit every decision. | | **Multi-provider support** | Anthropic, OpenAI, Google, DeepSeek, Ollama, Azure, Bedrock, Vertex. Switch providers with one config change. | From b8811c4546a0a24a38dd0700091b8d396f00cf5b Mon Sep 17 00:00:00 2001 From: HiranoMasaaki Date: Thu, 12 Feb 2026 18:29:58 +0900 Subject: [PATCH 2/2] Update: Align tagline with GitHub description MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit "The Agentic AI Runtime" / "The Agent Runtime" → "The Declarative Runtime for Agentic AI" Co-Authored-By: Claude Opus 4.6 --- README.md | 2 +- docs/README.md | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 427e5474..15e10682 100644 --- a/README.md +++ b/README.md @@ -1,4 +1,4 @@ -# Perstack: The Agentic AI Runtime +# Perstack: The Declarative Runtime for Agentic AI

npm version diff --git a/docs/README.md b/docs/README.md index 306a1cd5..16e9b671 100644 --- a/docs/README.md +++ b/docs/README.md @@ -1,5 +1,5 @@ --- -title: "Perstack: The Agent Runtime" +title: "Perstack: The Declarative Runtime for Agentic AI" --- Define AI agents as **Experts** in natural language. Execute them with deterministic, event-derived tracking. Each Expert runs in its own isolated context — no shared state, no prompt bloat, full execution history.