diff --git a/ai/honcho/config.toml b/ai/honcho/config.toml
index 7a6b986..2e4ddac 100644
--- a/ai/honcho/config.toml
+++ b/ai/honcho/config.toml
@@ -50,6 +50,7 @@ FLUSH_ENABLED = true
 transport = "openai"
 model = "deepseek-v4-flash"
 base_url = "https://opencode.ai/zen/go/v1"
+api_key_env = "OPENAI_API_KEY"
 
 # --- Dialectic ---
 [dialectic]
@@ -63,6 +64,7 @@ MAX_OUTPUT_TOKENS = 512
 transport = "openai"
 model = "deepseek-v4-flash"
 base_url = "https://opencode.ai/zen/go/v1"
+api_key_env = "OPENAI_API_KEY"
 
 [dialectic.levels.low]
 MAX_TOOL_ITERATIONS = 3
@@ -70,6 +72,7 @@ MAX_TOOL_ITERATIONS = 3
 transport = "openai"
 model = "deepseek-v4-flash"
 base_url = "https://opencode.ai/zen/go/v1"
+api_key_env = "OPENAI_API_KEY"
 
 [dialectic.levels.medium]
 MAX_TOOL_ITERATIONS = 2
@@ -77,6 +80,7 @@ MAX_TOOL_ITERATIONS = 2
 transport = "openai"
 model = "deepseek-v4-flash"
 base_url = "https://opencode.ai/zen/go/v1"
+api_key_env = "OPENAI_API_KEY"
 
 [dialectic.levels.high]
 MAX_TOOL_ITERATIONS = 4
@@ -84,6 +88,7 @@ MAX_TOOL_ITERATIONS = 4
 transport = "openai"
 model = "deepseek-v4-flash"
 base_url = "https://opencode.ai/zen/go/v1"
+api_key_env = "OPENAI_API_KEY"
 
 [dialectic.levels.max]
 MAX_TOOL_ITERATIONS = 10
@@ -91,6 +96,7 @@ MAX_TOOL_ITERATIONS = 10
 transport = "openai"
 model = "deepseek-v4-flash"
 base_url = "https://opencode.ai/zen/go/v1"
+api_key_env = "OPENAI_API_KEY"
 
 # --- Summary ---
 [summary]
@@ -102,6 +108,7 @@ MESSAGES_PER_LONG_SUMMARY = 60
 transport = "openai"
 model = "deepseek-v4-flash"
 base_url = "https://opencode.ai/zen/go/v1"
+api_key_env = "OPENAI_API_KEY"
 
 # --- Dream ---
 [dream]