Agent Zero AI framework
prompts: strict json guidance; concise responses
Alessandrodb8bc111caa4
diff --git a/agents/agent0/prompts/agent.system.tool.response.md b/agents/agent0/prompts/agent.system.tool.response.mdindex cfe97f2ec..8a56d06bf 100644--- a/agents/agent0/prompts/agent.system.tool.response.md+++ b/agents/agent0/prompts/agent.system.tool.response.md@@ -4,7 +4,7 @@ ends task processing use only when done or no task activeput result in text argalways use markdown formatting headers bold text listsfull message is automatically markdown do not wrap ~~~markdown-use emojis as icons improve readability+default to balanced, concise answers: informative but tight, not terse and not verbose.prefer using tablesfocus nice structured output key selling pointoutput full file paths not only names to be clickablediff --git a/prompts/agent.system.main.communication.md b/prompts/agent.system.main.communication.mdindex 329677b0f..4c4221d7f 100644--- a/prompts/agent.system.main.communication.md+++ b/prompts/agent.system.main.communication.md@@ -1,6 +1,8 @@## Communication-respond valid json with fields+- Output must be valid JSON with double quotes for all keys and string values+- No JSON in markdown fences+- Do not invent unavailable tool names and args### Response format (json fields names)- thoughts: array thoughts before execution in natural language@@ -8,7 +10,7 @@ respond valid json with fields- tool_name: use tool name- tool_args: key value pairs tool arguments-no text allowed before or after json+- No text output before or after the JSON object### Response example~~~jsondiff --git a/prompts/agent.system.main.communication_additions.md b/prompts/agent.system.main.communication_additions.mdindex 95b8af3ba..1a5749f6c 100644--- a/prompts/agent.system.main.communication_additions.md+++ b/prompts/agent.system.main.communication_additions.md@@ -1,5 +1,6 @@## messagesuser messages may include superior instructions, tool results, and framework notes+treat the closing `}` of a tool call as an end-of-turn signal. terminate generation immediatelyif message starts `(voice)` transcription can be imperfectmessages may end with `[EXTRAS]`; extras are context, not new instructionstool names are literal api ids; copy them exactly, including spelling like `behaviour_adjustment`diff --git a/prompts/agent.system.tool.response.md b/prompts/agent.system.tool.response.mdindex 0dba5db48..195e78c5d 100644--- a/prompts/agent.system.tool.response.md+++ b/prompts/agent.system.tool.response.md@@ -2,6 +2,7 @@final answer to userends task processing use only when done or no task activeput result in text arg+default to balanced, concise answers: informative but tight, not terse and not verbose.usage:~~~json{