TODOs (in priority order)

• Use custom click argument types to support the configlib for functions.
• Design a module for benchmark building pipelines
• Use reflection for automated logger configuration for the log module
• Implement log.watchdog
• Figure out how to strictly separate load_config and load_env for configurable classes
• Simplify the usage pattern of load_config and load_env
• Add error messages if using a configuration setting before it is loaded
• Develop a standard operating procedure (sop) for redacting the identity information for anonymous artifact submission in paper reviewing
• Refine the ergonomics of prompt
• Add format specifiers to prompt
• Add a py-tree-sitter wrapper module
• Move in the LLM pricing tracker
• Develop a sop to integrate Python code with native language code (such as Rust and Zig) to accommodate memory/time critical scenarios
• We should implement IntervalTree in the alg module in Zig, as the Python implementation consumes too much memory Deferred: The current status of Zig toolchains is immature
• subshell sometimes misses outputs, especially when PHDKIT_SIMPLE_SUBSHELL is on
• Consider how to use configlib for the context manager protocol
• Explore how to handle optional settings in configlib
• Add postload functionalities for configlib
• Add docs for default setting values in configlib
• Implement the subshell functionality of the rich wrapper module
• Make infix_fn type-safe
• Audit this whole "vibe-coding" interval tree implementation for correctness and performance
• Consider how to handle configurations of base classes in configlib
• Implement the dyntqdm functionality of the rich wrapper module Already supported by rich
• Test whether configlib handles arrays in TOML correctly
• Add a flag to subshell to switch it to a simple interface when interacting with coding agents