diff --git a/README.md b/README.md index 7648ded..1550b4a 100644 --- a/README.md +++ b/README.md @@ -10,19 +10,20 @@ For usage examples, see [doc/USAGE.md](https://github.com/lneely/ollie/blob/main ``` ollie/ - a/ dir: agent configs (r/w, backed by ~/.config/ollie/agents/) - .json r/w: agent config; supports mv, cp, rm - backends read: list of ollie-provided backends - help read: help file (backed by ~/.config/ollie/help.md) - p/ dir: prompt templates (read, backed by ~/.config/ollie/prompts/) - 00_base.md read: tone, accuracy, task execution, environment - 01_ollie.md read: session identity, filesystem layout, session lifecycle - 02_reasoning.md read: reasoning instructions - 03_memory.md read: memory instructions - 04_task.md read: task planning instructions - 05_edit-text.md read: text editing instructions - 06_skills.md read: skill discovery and loading - 07_tools.md read: tool script usage + a/ dir: agent configs (r/w, backed by ~/.config/ollie/agents/) + .json r/w: agent config; supports mv, cp, rm + backends read: list of ollie-provided backends + help read: help file (backed by ~/.config/ollie/help.md) + p/ dir: prompt templates (read, backed by ~/.config/ollie/prompts/) + 00_base.md read: tone, accuracy, task execution, environment + 01_ollie.md read: session identity, filesystem layout, session lifecycle + 02_skills.md read: skill discovery and loading + 03_tool-files.md read: text editing instructions + 04_tool-reasoning.md read: reasoning instructions + 05_tool-memory.md read: memory instructions + 06_tool-plan.md read: task planning instructions + 07_tool-subagents.md read: subagent usage + 08_tool-elevate.md read: controlled escape from execute_code sandbox Prompt files are assembled in lexical order and rendered as Go templates at session start. Changes take effect the next time the system prompt is rebuilt: on new session @@ -70,7 +71,7 @@ ollie/