[BUG] Conditional file loading with @? syntax not working - language-specific guidelines not auto-loaded

Environment:

• Claude Code version: [current version from 1.0.69 (Claude Code)]
• OS: macOS (Darwin 24.5.0)
• Working directory: Go project with go.mod file present

Description:
The conditional loading feature for language-specific guidelines is not working as documented. According to the documentation, files should be automatically loaded based on
project type detection, but this is not happening.

Expected Behavior:
When working in a Go project (with go.mod present), the @?GO.md conditional loading should automatically include ~/.claude/GO.md in the context, as shown by /status command
output.

Actual Behavior:
The /status command shows that GO.md is not loaded despite:

1. ~/.claude/GO.md file exists and is readable
2. go.mod file exists in working directory
3. Conditional loading syntax is properly configured in ~/.claude/CLAUDE.md

Steps to Reproduce:

1. Create a ~/.claude/CLAUDE.md file with conditional loading syntax:

  @?GO.md # Load only if Go project detected (go.mod exists)

1. Ensure ~/.claude/GO.md exists
2. Navigate to a directory containing go.mod
3. Start Claude Code session
4. Run /status command
5. Observe that GO.md is not listed in the Memory section

Current Configuration:
# ~/.claude/CLAUDE.md excerpt
@?GO.md # Load only if Go project detected (go.mod exists)
@?NODE.md # Load only if Node.js project detected (package.json exists)
@?PYTHON.md # Load only if Python project detected (requirements.txt/pyproject.toml exists)
@?RUST.md # Load only if Rust project detected (Cargo.toml exists)
@?JAVA.md # Load only if Java project detected (pom.xml/build.gradle exists)

Actual /status output:
Memory • /memory
L user: ~/.claude/CLAUDE.md

L project: CLAUDE.md

Impact:
Language-specific development guidelines are not automatically available, requiring manual workarounds and reducing the effectiveness of project-specific configuration.

Possible Causes:

1. @? conditional syntax not implemented
2. File detection logic not working from current working directory
3. Case sensitivity issues in file detection
4. Documentation ahead of implementation

Workaround:
Currently using direct file reads to access language-specific guidelines manually.