update docs

2 files changed, 35 insertions, 6 deletions

diff --git a/README.md b/README.md
index 7fb9bf3..7284e90 100644
--- a/README.md
+++ b/README.md
@@ -18,16 +18,11 @@ It has got improved capabilities for Go code understanding (for example, create
 
 ## Documentation
 
-* [Configuration guide](docs/configuration.md)
 * [Build and install guide](docs/buildandinstall.md)
+* [Configuration guide](docs/configuration.md)
 * [Usage examples](docs/usage.md)
 * [Helix + tmux quickstart](docs/tmux.md)
 
-## Hexai CLI
-
-- When invoked without arguments, `hexai` opens your editor (`$HEXAI_EDITOR` or `$EDITOR`) on a temporary `.md` file to capture a prompt, and combines it with any piped stdin.
-- With arguments, behavior is unchanged (no editor spawn).
-
 ## Tmux Status Line
 
 See the [tmux integration guide](docs/tmux.md) for details on configuring the status line.
diff --git a/SCRATCHPAD.md b/SCRATCHPAD.md
new file mode 100644
index 0000000..afea5aa
--- /dev/null
+++ b/SCRATCHPAD.md
@@ -0,0 +1,34 @@
+# Project scratch pad
+
+This document shows future items and items in progress. Already completed ones are deleted from this document as updates occur.
+
+## Features
+
+* [ ] Keep global stats about LLM usage for the tmux pane!
+* [ ] No a feature, but verify my OpenAI API account so I can use GPT-5 via the API.
+* [ ] In-editor chat triggers should be context aware of the current file, buffer and function!
+* [ ] Kagi FastGPT for in-editor search
+  - Think about an in-editor chat trigger, maybe with S> for search!
+* [ ] Test whethe GitHub Copilot support actually works now, and if not, fix it!
+* [ ] Be able to re-configure the temperature in-editor
+* [ ] For in-editor chat add a way to print current hexai status such as
+  * active LLM
+  * active stats
+  * current config printed out
+  * could use a special keyword like /status?TRIGGER (e.g. > as TRIGGER)
+  * what other slasm commands could we think of?
+* [ ] Be able to switch LLMs ad-hoc.
+
+## More
+
+* [ ] Configure hexai-lsp multiple times, but with different LLM backends? E.g. a remote cloud one and a local one?
+* [/] Review documentation
+* [/] Manual review the code 
+* [ ] Useful: https://deepwiki.com/helix-editor/helix/4.3-language-server-protocol
+
+## Additional Ideas (Nice to Have)
+
+*   **LSP: Inlay Hints (`textDocument/inlayHint`)**: Use AI to provide dynamic, inline hints like inferred types, performance warnings (e.g., `O(n^2)`), or security alerts.
+*   **LSP: Hover (`textDocument/hover`)**: Enrich hover popups with AI-generated natural language explanations of code, usage examples, and complexity analysis.
+*   **LSP: Semantic Tokens (`textDocument/semanticTokens`)**: Implement semantic tokenization to build a deeper understanding of the code, which would improve the accuracy and context-awareness of all other AI features.
+*   **LSP: Rename (`textDocument/rename`)**: Add safe rename capabilities, potentially enhanced with AI to proactively suggest better, more descriptive names for variables and functions.