Agenta-AI · mmabrouk · Jul 25, 2026 · Jul 30, 2026 · Jul 30, 2026 · Jul 31, 2026
diff --git a/docs/blog/entries/agenta-is-now-a-workspace-for-building-agents.mdx b/docs/blog/entries/agenta-is-now-a-workspace-for-building-agents.mdx
@@ -0,0 +1,83 @@
+---
+title: "Agenta Is Now a Workspace for Building Agents"
+slug: agenta-is-now-a-workspace-for-building-agents
+date: 2026-07-21
+tags: [v0.105.7]
+description: "Agenta is now a workspace for building and running agents: assistants made of instructions, tools, skills, permissions, and files that you can chat with or run in the background."
+---
+
+<Summary>
+
+<iframe
+  width="100%"
+  height="400"
+  src="https://www.youtube.com/embed/Y5l2BPRkKC8"
+  title="Agenta Is Now a Workspace for Building Agents"
+  frameBorder="0"
+  allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
+  allowFullScreen
+></iframe>
+
+Agenta is now a workspace for building and running agents, not only a platform for managing prompts. You build an agent by chatting with it: describe the job, connect the apps it needs, and correct it through feedback until it works the way you want.
+
+An agent is made of instructions, tools, skills, permissions, and files. You can work with it in chat, or run it in the background on a schedule or when something happens in an app you connected. Agenta runs agents on Claude Code or Pi, records a version every time the configuration changes, and traces every run.
+
+</Summary>
+
+{/* truncate */}
+
+Agenta started as a platform for managing and evaluating prompts. With this release, it becomes a workspace where you build agents: assistants that plan their own steps, call tools, and get work done, rather than answer a single prompt.
+
+<iframe
+  width="100%"
+  height="400"
+  src="https://www.youtube.com/embed/Y5l2BPRkKC8"
+  title="Agenta Is Now a Workspace for Building Agents"
+  frameBorder="0"
+  allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
+  allowFullScreen
+></iframe>
+
+## What an agent is made of
+
+An agent starts empty. You build it for a job by chatting with it, and the agent can write its own configuration as you talk. That configuration has five parts.
+
+**Instructions** describe the job: what the agent does, who it serves, and how it should report back.
+
+**Tools** are actions the agent can take in other systems, like creating a GitHub issue or sending a Slack message. Tools come from Agenta's built-in integrations, which cover mail, chat, calendars, documents, issue trackers, and CRMs, or from MCP servers you connect.
+
+**Skills** are short handbooks for tasks the agent does sometimes. The agent loads a skill only when the task calls for it, so it isn't carrying instructions for every possible task on every turn.
+
+**Permissions** decide which actions the agent takes on its own and which wait for your approval. You set a default for the agent, such as "read on its own, ask before writing," and you can override that default for individual tools.
+
+**Files** live in the agent's folder: the drafts it writes, the notes it keeps, and the reference material you give it. You and the agent share this folder, so you can write documents or maintain a wiki together.
+
+See [Agents](/concepts/agents) for the full picture.
+
+## Two ways to use an agent
+
+You can work with an agent in chat. You give it a task, it does the work, and you review the result together.
+
+You can also run an agent as an automation, with nobody in the chat. An automation starts on a schedule, such as every weekday morning, or when something happens in an app you connected, such as a new row in a sheet or a new issue. The agent does the work and leaves the result somewhere you'll see it.
+
+## Choose your harness
+
+A harness is the program that drives the model: it sends the conversation to the model, runs the tool calls the model asks for, and feeds the results back. Agenta ships two harnesses, Claude Code and Pi, and you can switch harness or model without rebuilding the agent.
+
+If you self-host Agenta, an agent can also run against a Claude or ChatGPT subscription you already pay for, instead of moving every task to metered API billing.
+
+## Versions and traces
+
+Every change to an agent's configuration is recorded as a version. You can read back what changed, when, and why, and compare versions to understand why an agent behaves differently than it did before.
+
+Agenta also traces every run. You can inspect every model and tool call, and track requests, token usage, and estimated cost for each agent.
+
+## Open source
+
+Agenta is open source under the MIT license. You can inspect the code, self-host it, and run it in commercial projects.
+
+## Getting started
+
+- [Quick start](/getting-started/quick-start) gets you a working agent.
+- [Agents](/concepts/agents) explains what an agent is made of.
+- [Build your first agent](/learn/build-your-first-agent) walks through the whole path, from an empty agent to one with tools and skills.
diff --git a/docs/blog/entries/annotate-your-llm-response-preview.mdx b/docs/blog/entries/annotate-your-llm-response-preview.mdx
@@ -20,7 +20,7 @@ This is useful to:
 - Run custom evaluation workflows
 - Measure application performance in real-time
 
-Check out the how to [annotate traces from API](/observability/trace-with-python-sdk/annotate-traces) for more details. Or try our new tutorial (available as [jupyter notebook](https://github.com/Agenta-AI/agenta/blob/main/examples/jupyter/capture_user_feedback.ipynb)) [here](/tutorials/cookbooks/capture-user-feedback).
+Check out the how to [annotate traces from API](/1.0/observability/trace-with-python-sdk/annotate-traces) for more details. Or try our new tutorial (available as [jupyter notebook](https://github.com/Agenta-AI/agenta/blob/main/examples/jupyter/capture_user_feedback.ipynb)) [here](/1.0/tutorials/cookbooks/capture-user-feedback).
 
 <Image
   style={{
@@ -50,7 +50,7 @@ This is useful to:
 - Run custom evaluation workflows
 - Measure application performance in real-time
 
-Check out the how to [annotate traces from API](/observability/trace-with-python-sdk/annotate-traces) for more details. Or try our new tutorial (available as [jupyter notebook](https://github.com/Agenta-AI/agenta/blob/main/examples/jupyter/observability/capture_user_feedback.ipynb)) [here](/tutorials/cookbooks/capture-user-feedback).
+Check out the how to [annotate traces from API](/1.0/observability/trace-with-python-sdk/annotate-traces) for more details. Or try our new tutorial (available as [jupyter notebook](https://github.com/Agenta-AI/agenta/blob/main/examples/jupyter/observability/capture_user_feedback.ipynb)) [here](/1.0/tutorials/cookbooks/capture-user-feedback).
 
 <Image
   style={{

diff --git a/docs/blog/entries/chat-sessions-observability.mdx b/docs/blog/entries/chat-sessions-observability.mdx
@@ -148,9 +148,9 @@ Identify latency issues across entire conversations, not just single requests. S
 
 Learn more in our documentation:
 
-- [Track Chat Sessions (Python SDK)](/observability/trace-with-python-sdk/track-chat-sessions)
-- [Session Tracking (OpenTelemetry)](/observability/trace-with-opentelemetry/session-tracking)
-- [Observability Overview](/observability/overview)
+- [Track Chat Sessions (Python SDK)](/1.0/observability/trace-with-python-sdk/track-chat-sessions)
+- [Session Tracking (OpenTelemetry)](/1.0/observability/trace-with-opentelemetry/session-tracking)
+- [Observability Overview](/1.0/observability/overview)
 
 ## What's Next
 

diff --git a/docs/blog/entries/customize-llm-as-a-judge-output-schemas.mdx b/docs/blog/entries/customize-llm-as-a-judge-output-schemas.mdx
@@ -27,7 +27,7 @@ The LLM-as-a-Judge evaluator now supports custom output schemas. Create multiple
 
 You can configure output types (binary, multiclass), include reasoning to improve prediction quality, or provide a raw JSON schema with any structure you define. Use these custom schemas in your evaluations to capture exactly the feedback you need.
 
-Learn more in the [LLM-as-a-Judge documentation](/evaluation/configure-evaluators/llm-as-a-judge).
+Learn more in the [LLM-as-a-Judge documentation](/1.0/evaluation/configure-evaluators/llm-as-a-judge).
 
 </Summary>
 
@@ -93,4 +93,4 @@ To use custom output schemas with LLM-as-a-Judge:
 4. For advanced use, provide your JSON schema
 5. Run your evaluation
 
-Learn more in the [LLM-as-a-Judge documentation](/evaluation/configure-evaluators/llm-as-a-judge).
+Learn more in the [LLM-as-a-Judge documentation](/1.0/evaluation/configure-evaluators/llm-as-a-judge).
diff --git a/docs/blog/entries/deployment-webhooks-and-github-automations.mdx b/docs/blog/entries/deployment-webhooks-and-github-automations.mdx
@@ -12,7 +12,7 @@ You can now trigger automations when a prompt deployment happens in Agenta. Send
 
 This makes it easier to connect prompt deployments to CI, repository sync jobs, and pull request workflows. If your GitHub workflow needs the latest prompt content, fetch it from Agenta during the run and commit the result back to your repo.
 
-Learn more: [Webhooks](/prompt-engineering/integrating-prompts/webhooks) | [GitHub Automations](/prompt-engineering/integrating-prompts/github)
+Learn more: [Webhooks](/1.0/prompt-engineering/integrating-prompts/webhooks) | [GitHub Automations](/1.0/prompt-engineering/integrating-prompts/github)
 
 </Summary>
 
@@ -56,6 +56,6 @@ For GitHub automations, Agenta calls the GitHub API directly. `repository_dispat
 
 Start with these guides:
 
-- [Webhooks](/prompt-engineering/integrating-prompts/webhooks)
-- [GitHub](/prompt-engineering/integrating-prompts/github)
-- [Fetch Prompts via SDK/API](/prompt-engineering/integrating-prompts/fetch-prompt-programatically)
+- [Webhooks](/1.0/prompt-engineering/integrating-prompts/webhooks)
+- [GitHub](/1.0/prompt-engineering/integrating-prompts/github)
+- [Fetch Prompts via SDK/API](/1.0/prompt-engineering/integrating-prompts/fetch-prompt-programatically)
diff --git a/docs/blog/entries/documentation-architecture-overhaul.mdx b/docs/blog/entries/documentation-architecture-overhaul.mdx
@@ -15,8 +15,8 @@ We've completely rewritten and restructured our documentation with a new archite
 
 Key improvements include:
 - **[Diataxis Framework](https://diataxis.fr/)**: Organized content into Tutorials, How-to Guides, Reference, and Explanation sections for better discoverability
-- **[Expanded Observability Docs](/observability/overview)**: Added missing documentation for tracing, annotations, and observability features
-- **[JavaScript/TypeScript Support](/observability/quick-start-opentelemetry)**: Added code examples and documentation for JavaScript developers alongside Python
+- **[Expanded Observability Docs](/1.0/observability/overview)**: Added missing documentation for tracing, annotations, and observability features
+- **[JavaScript/TypeScript Support](/1.0/observability/quick-start-opentelemetry)**: Added code examples and documentation for JavaScript developers alongside Python
 - **Ask AI Feature**: Ask questions directly to the documentation for instant answers
 
 </Summary>
@@ -33,10 +33,10 @@ We've reorganized all documentation using the [Diataxis framework](https://diata
 
 One of the biggest gaps in our previous documentation was observability. We've added comprehensive documentation covering:
 
-- [Tracing with OpenTelemetry](/observability/trace-with-opentelemetry/getting-started)
-- [Tracing LLM applications with JS/TS](/observability/quick-start-opentelemetry)
-- [Using the Metrics API to fetch metrics](/observability/query-data/analytics-data)
-- [Using the Query API to fetch traces](/observability/query-data/query-api)
+- [Tracing with OpenTelemetry](/1.0/observability/trace-with-opentelemetry/getting-started)
+- [Tracing LLM applications with JS/TS](/1.0/observability/quick-start-opentelemetry)
+- [Using the Metrics API to fetch metrics](/1.0/observability/query-data/analytics-data)
+- [Using the Query API to fetch traces](/1.0/observability/query-data/query-api)
 
 ### JavaScript/TypeScript Support
 

diff --git a/docs/blog/entries/documentation-overhaul-new-models-and-platform-improvements.mdx b/docs/blog/entries/documentation-overhaul-new-models-and-platform-improvements.mdx
@@ -13,7 +13,7 @@ We've made significant improvements across Agenta with a major documentation ove
 
 We've completely rewritten our prompt management and prompt engineering documentation.
 
-Start exploring the new documentation in our updated [Quick Start Guide](/prompt-engineering/quick-start).
+Start exploring the new documentation in our updated [Quick Start Guide](/1.0/prompt-engineering/quick-start).
 
 **New Model Support**:
 
@@ -55,7 +55,7 @@ We've made significant improvements across Agenta with a major documentation ove
 
 We've completely rewritten our prompt management and prompt engineering documentation.
 
-Start exploring the new documentation in our updated [Quick Start Guide](/prompt-engineering/quick-start).
+Start exploring the new documentation in our updated [Quick Start Guide](/1.0/prompt-engineering/quick-start).
 
 **New Model Support**:
 

diff --git a/docs/blog/entries/dspy-integration.mdx b/docs/blog/entries/dspy-integration.mdx
@@ -37,4 +37,4 @@ We've added DSPy integration to Agenta. You can now trace and debug your DSPy ap
   loading="lazy"
 />
 
-[**View the full DSPy integration →**](/integrations/frameworks/dspy/observability)
+[**View the full DSPy integration →**](/1.0/integrations/frameworks/dspy/observability)
diff --git a/docs/blog/entries/evaluation-sdk.mdx b/docs/blog/entries/evaluation-sdk.mdx
@@ -24,7 +24,7 @@ You can now run programmatic evaluations of complex AI agents and workflows dire
 
 The SDK lets you create test sets in code or fetch them from Agenta. You can use built-in evaluators like LLM-as-a-Judge, semantic similarity, or regex matching. You can also write custom Python evaluators. The SDK evaluates end-to-end workflows or specific spans in execution traces. Evaluations run on your own infrastructure; results display in the Agenta dashboard.
 
-Check out the [Evaluation SDK documentation](/evaluation/evaluation-from-sdk/quick-start) to get started.
+Check out the [Evaluation SDK documentation](/1.0/evaluation/evaluation-from-sdk/quick-start) to get started.
 
 </Summary>
 
@@ -122,4 +122,4 @@ Every evaluation run gets a shareable dashboard link. The dashboard shows full e
 
 ## Next Steps
 
-Check out the [Quick Start Guide](/evaluation/evaluation-from-sdk/quick-start) to build your first evaluation.
+Check out the [Quick Start Guide](/1.0/evaluation/evaluation-from-sdk/quick-start) to build your first evaluation.
diff --git a/docs/blog/entries/filtering-traces-by-annotation.mdx b/docs/blog/entries/filtering-traces-by-annotation.mdx
@@ -19,7 +19,7 @@ The new annotation filtering lets you find:
 - Spans evaluated by a specific evaluator
 - Spans with user feedback like `success=True`
 
-This enables powerful workflows: [capture user feedback](/tutorials/cookbooks/capture-user-feedback) from your app, filter to find traces with bad feedback, add them to test sets, and improve your prompts based on real user data.
+This enables powerful workflows: [capture user feedback](/1.0/tutorials/cookbooks/capture-user-feedback) from your app, filter to find traces with bad feedback, add them to test sets, and improve your prompts based on real user data.
 
 </Summary>
 
@@ -47,7 +47,7 @@ Filter traces based on evaluations and feedback:
 
 This feature enables powerful workflows:
 
-1. **Capture user feedback** from your application using our API ([see tutorial](/tutorials/cookbooks/capture-user-feedback))
+1. **Capture user feedback** from your application using our API ([see tutorial](/1.0/tutorials/cookbooks/capture-user-feedback))
 2. **Filter traces** to find those with bad feedback or low scores
 3. **Add them to test sets** to track problematic cases
 4. **Improve your prompts** based on real user feedback

diff --git a/docs/blog/entries/jinja2-template-support.mdx b/docs/blog/entries/jinja2-template-support.mdx
@@ -13,7 +13,7 @@ import Image from "@theme/IdealImage";
 
 You can now use Jinja2 templates in your prompts. Jinja2 is available in both the Playground and in prompt management.
 
-Learn more in our [blog post](https://agenta.ai/blog/launch-week-2-day-5-jinja2-prompt-templates) or check the [documentation](/prompt-engineering/playground/using-playground#switching-template-formats).
+Learn more in our [blog post](https://agenta.ai/blog/launch-week-2-day-5-jinja2-prompt-templates) or check the [documentation](/1.0/prompt-engineering/playground/using-playground#switching-template-formats).
 
 </Summary>
 
@@ -23,7 +23,7 @@ We're excited to announce a powerful update to the Agenta playground. You can no
 
 This means you can add sophisticated logic directly into your prompt templates. Use conditional statements, apply filters to variables, and transform data on the fly.
 
-Learn more in our [blog post](https://agenta.ai/blog/launch-week-2-day-5-jinja2-prompt-templates) or check the [documentation](/prompt-engineering/playground/using-playground#switching-template-formats).
+Learn more in our [blog post](https://agenta.ai/blog/launch-week-2-day-5-jinja2-prompt-templates) or check the [documentation](/1.0/prompt-engineering/playground/using-playground#switching-template-formats).
 
 ## Example
 

diff --git a/docs/blog/entries/json-multi-field-match.mdx b/docs/blog/entries/json-multi-field-match.mdx
@@ -129,4 +129,4 @@ Existing JSON Field Match configurations continue to work. We recommend migratin
 
 ## Next Steps
 
-Learn more about configuring and using the JSON Multi-Field Match evaluator in the [Classification and Entity Extraction Evaluators](/evaluation/configure-evaluators/classification-entity-extraction#json-multi-field-match) documentation.
+Learn more about configuring and using the JSON Multi-Field Match evaluator in the [Classification and Entity Extraction Evaluators](/1.0/evaluation/configure-evaluators/classification-entity-extraction#json-multi-field-match) documentation.
diff --git a/docs/blog/entries/llamaindex-integration.mdx b/docs/blog/entries/llamaindex-integration.mdx
@@ -54,4 +54,4 @@ The integration is auto-instrumentation - just add one line of code and you'll s
 
 This helps when you need to understand what's happening inside your RAG pipeline, track performance bottlenecks, or debug issues in production.
 
-Check out the [tutorial](/integrations/frameworks/llamaindex/observability) and the [Jupyter notebook](https://github.com/Agenta-AI/agenta/blob/main/examples/jupyter/llamaindex_integration.ipynb) for more details.
+Check out the [tutorial](/1.0/integrations/frameworks/llamaindex/observability) and the [Jupyter notebook](https://github.com/Agenta-AI/agenta/blob/main/examples/jupyter/llamaindex_integration.ipynb) for more details.
diff --git a/docs/blog/entries/multiple-metrics-in-human-evaluation.mdx b/docs/blog/entries/multiple-metrics-in-human-evaluation.mdx
@@ -11,7 +11,7 @@ We rebuilt the human evaluation workflow from scratch. Now you can set multiple
 
 This lets you evaluate the same output on different metrics like **relevance** or **completeness**. You can also create binary, numerical scores, or even use strings for **comments** or **expected answer**. 
 
-Watch the video below and read the [post](/changelog/multiple-metrics-in-human-evaluation) for more details. Or check out the [docs](/evaluation/human-evaluation/quick-start) to learn how to use the new human evaluation workflow.
+Watch the video below and read the [post](/changelog/multiple-metrics-in-human-evaluation) for more details. Or check out the [docs](/1.0/evaluation/human-evaluation/quick-start) to learn how to use the new human evaluation workflow.
 
 <div style={{display: 'flex', justifyContent: 'center', marginTop: "20px", marginBottom: "20px", flexDirection: 'column', alignItems: 'center'}}>
   <iframe
@@ -46,6 +46,6 @@ This unlocks a whole new set of use cases:
 - Use human evaluation to bootstrap automatic evaluation. You can annotate your outputs with the expected answer or a rubic, then use it to set up an automatic evaluation.
 
 
-Watch the video below and read the [post](/changelog/multiple-metrics-in-human-evaluation) for more details. Or check out the [docs](/evaluation/human-evaluation/quick-start) to learn how to use the new human evaluation workflow.
+Watch the video below and read the [post](/changelog/multiple-metrics-in-human-evaluation) for more details. Or check out the [docs](/1.0/evaluation/human-evaluation/quick-start) to learn how to use the new human evaluation workflow.
 
 ---
Original file line number	Diff line number	Diff line change
Expand Up		@@ -129,4 +129,4 @@ Existing JSON Field Match configurations continue to work. We recommend migratin

		## Next Steps

		Learn more about configuring and using the JSON Multi-Field Match evaluator in the [Classification and Entity Extraction Evaluators](/evaluation/configure-evaluators/classification-entity-extraction#json-multi-field-match) documentation.
		Learn more about configuring and using the JSON Multi-Field Match evaluator in the [Classification and Entity Extraction Evaluators](/1.0/evaluation/configure-evaluators/classification-entity-extraction#json-multi-field-match) documentation.
Original file line number	Diff line number	Diff line change
Expand Up		@@ -54,4 +54,4 @@ The integration is auto-instrumentation - just add one line of code and you'll s

		This helps when you need to understand what's happening inside your RAG pipeline, track performance bottlenecks, or debug issues in production.

		Check out the [tutorial](/integrations/frameworks/llamaindex/observability) and the [Jupyter notebook](https://github.com/Agenta-AI/agenta/blob/main/examples/jupyter/llamaindex_integration.ipynb) for more details.
		Check out the [tutorial](/1.0/integrations/frameworks/llamaindex/observability) and the [Jupyter notebook](https://github.com/Agenta-AI/agenta/blob/main/examples/jupyter/llamaindex_integration.ipynb) for more details.