diff --git a/README.md b/README.md
index 54a5e0f..1639cd3 100644
--- a/README.md
+++ b/README.md
@@ -21,6 +21,9 @@ While creating an agent prototype is easy, increasing its reliability and robust
   - Calculate the cost of LLM calls and agent runs
   - Stop agent runs when certain criteria are met, e.g. to limit the number of steps
   - Use several different LLM models in one agent
+- Agent HTTP Server
+  - Agent runs can be started, stopped, and observed via HTTP API
+  - Can host multiple agents
 - Supported LLM models
   - OpenAI text completion models (`text-davinci-003` etc.)
   - OpenAI chat completion models (`gpt-4`, `gpt-3.5-turbo`)
@@ -68,7 +71,7 @@ Used features: `gpt-4`, tool execution in Docker container, agent with fixed set
 
 JS Agent implementation of [BabyAGI](https://github.com/yoheinakajima/babyagi).
 
-Features used: text completion model (`text-davinci-003`), customized console output, update tasks planning loop
+Features used: HTTP Agent server, text completion model (`text-davinci-003`), customized console output, update tasks planning loop
 
 ### [PDF Summarizer](https://github.com/lgrammel/js-agent/tree/main/examples/pdf-summarizer)
 
diff --git a/examples/babyagi/README.md b/examples/babyagi/README.md
index 6c4f44a..aca3e6f 100644
--- a/examples/babyagi/README.md
+++ b/examples/babyagi/README.md
@@ -61,7 +61,3 @@ curl -X POST -H "Content-Type: application/json" -d '{"objective":"solve world h
 ```bash
 ❯ curl -X POST -H "Content-Type: application/json" -d '{"reason": "need to shut down computer"}' http://127.0.0.1:30800/agent/babyagi/run/bxynsv4USkGoawtCYhte-w-0/cancel
 ```
-
-## Source Code
-
-See `agent/babyagi/agent.ts`.
diff --git a/packages/agent/README.md b/packages/agent/README.md
index 6e38957..516fb08 100644
--- a/packages/agent/README.md
+++ b/packages/agent/README.md
@@ -19,6 +19,9 @@ While creating an agent prototype is easy, increasing its reliability and robust
   - Calculate the cost of LLM calls and agent runs
   - Stop agent runs when certain criteria are met, e.g. to limit the number of steps
   - Use several different LLM models in one agent
+- Agent HTTP Server
+  - Agent runs can be started, stopped, and observed via HTTP API
+  - Can host multiple agents
 - Supported LLM models
   - OpenAI text completion models (`text-davinci-003` etc.)
   - OpenAI chat completion models (`gpt-4`, `gpt-3.5-turbo`)
@@ -66,7 +69,7 @@ Used features: `gpt-4`, tool execution in Docker container, agent with fixed set
 
 JS Agent implementation of [BabyAGI](https://github.com/yoheinakajima/babyagi).
 
-Features used: text completion model (`text-davinci-003`), customized console output, update tasks planning loop
+Features used: HTTP Agent server, text completion model (`text-davinci-003`), customized console output, update tasks planning loop
 
 ### [PDF Summarizer](https://github.com/lgrammel/js-agent/tree/main/examples/pdf-summarizer)