remy-index

Files (44)

• README.md
• README_zh.md
• SKILL.md
• bootstrap.py
• constants.py
• default_logic_config.template
• impact.py
• index_state.py
• llm_client.py
• llm_judge.py
• migrations.py
• __init__.py · parsers/
• base.py · parsers/
• c_cpp_parser.py · parsers/
• python_parser.py · parsers/
• registry.py · parsers/
• rust_parser.py · parsers/
• ts_parser.py · parsers/
• judge_propagation.md · prompts/
• summarize_cluster.md · prompts/
• summarize_file.md · prompts/
• summarize_symbol_c.md · prompts/
• summarize_symbol_python.md · prompts/
• summarize_symbol_rust.md · prompts/
• summarize_symbol_ts.md · prompts/
• propagation.py
• retrieval_projection.py
• run.py
• scanner.py
• schema.py
• schema.sql
• struct_scan.py
• summarizer.py
• symbol_names.py
• symbol_selection.py
• __init__.py · synthesizers/
• c_fnptr_dispatch.py · synthesizers/
• EXTENSION_PLAN.md · synthesizers/c_fnptr_profiles/
• SCHEMA.md · synthesizers/c_fnptr_profiles/
• __init__.py · synthesizers/c_fnptr_profiles/
• tee.py · synthesizers/c_fnptr_profiles/
• event_emitter.py · synthesizers/
• interface_dispatch.py · synthesizers/
• rust_trait_dispatch.py · synthesizers/

SKILL.md (first lines)

# Update Logic Index

Updates the semantic understanding of the codebase by scanning source files (Python, C, C++, TypeScript, Rust), extracting symbol signatures, and generating incremental summaries using an OpenAI-compatible LLM API.

Supports multi-language projects. Language is detected automatically by file extension:
- Python: `.py`
- C: `.c`, `.h`
- C++: `.cpp`, `.hpp`, `.cc`, `.cxx`, `.hh`, `.hxx`
- TypeScript: `.ts`, `.tsx`
- Rust: `.rs`

C/C++ and TypeScript/TSX parsing use regex-based extraction by default (zero external dependencies). Rust parsing requires the tree-sitter grammar (no regex fallback; without it `.rs` files fail scanning and keep previously indexed rows). Rust `impl Trait for Type` blocks are recorded as per-file `rust_trait_impl` facts; the global postprocess derives struct/enum `bases` from them (cross-file impls included, ambiguous short-name targets merge nothing) and synthesizes `trait-impl` edges at the impl site. Install `tree-sitter` for higher C/C++/TS precision and for Rust support:
```bash
pip install tree-sitter tree-sitter-c tree-sitter-cpp tree-sitter-typescript tree-sitter-rust
```

## Process

You MUST execute the following steps strictly in order.

### Phase 1: Check Configuration
1.  Check if `.claude/logic_index_config` exists.
2.  **If missing**:
    - Create it from `~/.claude/skills/remy-index/default_logic_config.template`.
    - Use `AskUserQuestion` to prompt: "Configuration generated. Please review `.claude/logic_index_config` to exclude unnecessary directories (to save Tokens). Continue?"
    - Wait for "Yes".
3.  **If exists**:
    - Read the file content.
    - Use `AskUserQuestion` to prompt: "Existing configuration found. The scan will run based on these rules (consuming Tokens). Proceed?"
    - Wait for "Yes".

### Phase 2: Execute Scanning
1.  Check if `.claude/logic_index.db` exists.
    - If **MISSING**: Output a first-run notice in the language specified by the loaded `language.md` directive.
2.  Execute the Python indexer:
    ```bash
    python "~/.claude/skills/remy-index/run.py"
    ```
3.  Wait for completion and inspect the exit code:
    - `0` (`success`): continue normally.
    - `2` (`partial`): report the incomplete stages, then continue to Phase 3 so any bootstrap confirmation marker is still handled.
    - `1` (`failed`): stop. Do not run injection; dirty paths remain queued for retry.

### Phase 2.5: Symbol Summary Confirmation (Trigger: run.py stdout contains SYMBOL_PENDING_CONFIRMATION)

After Phase 2, inspect the stdout produced by `run.py`:

1.  **Search the captured stdout for the line `SYMBOL_PENDING_CONFIRMATION`.** This line is emitted only when the effective symbol mode is `ask` (`REMY_SYMBOL_SUMMARY_MODE=ask`, or downgraded from `auto` because the pending-symbol count exceeds `REMY_SYMBOL_AUTO_SIZE_GUARD`). When this line is present, run.py has skipped the symbol layer, the file/cluster bootstrap, and the propagation pass; the call graph is already up to date.
2.  **If the line is absent**: Skip Phase 2.5 entirely.
3.  **If the line is present**: Parse the `pending_symbols=X` field, then use `AskUserQuestion` in the language specified by the loaded `language.md` directive:
    - Question (zh-CN): "检测到 {X} 个符号尚未生成 LLM 摘要。立即生成？（消耗 LLM tokens，随后自动继续 file/cluster 摘要与传播判定）"
    - Question (en): "{X} symbol(s) lack LLM summaries. Generate now? (consumes LLM tokens; file/cluster summaries and propagation follow automatically)"
    - Options:
        - "Yes — generate now (推荐)" — proceed to step 4.
        - "Skip this session" — proceed to step 5.
        - "Disable symbol summaries permanently" — proceed to step 6.
4.  **If user chose "Yes"**: Rerun the full pipeline, overriding the symbol mode to `auto`:
    ```bash
    python "~/.claude/skills/remy-index/run.py" --symbol-mode auto
    ```
    Wait for completion (the incremental structural scan is cheap; pending symbols resume automatically), then re-enter Phase 3 with the new stdout.
5.  **If user chose "Skip"**: Output a single notice: "Skipped. Summaries stay pending; run `/remy-index` again later to resume." Then continue to Phase 4 (the graph is usable without summaries).
6.  **If user chose "Disable permanently"**: Output the command for the user to apply manually: "Set `REMY_SYMBOL_SUMMARY_MODE=never` through `remy-cc config --path <project>`." Then continue to Phase 4.

### Phase 3: Hierarchical Bootstrap Confirmation (Trigger: run.py stdout contains BOOTSTRAP_PENDING_CONFIRMATION)

After Phase 2, inspect the stdout produced by `run.py` and decide whether to ask the user about generating file/cluster summaries:

1.  **Search the captured stdout for the line `BOOTSTRAP_PENDING_CONFIRMATION`.** This line is emitted only when `REMY_SUMMARY_BOOTSTRAP_MODE=ask` (explicit or downgraded from `auto` due to a missing API key or file count exceeding `REMY_BOOTSTRAP_AUTO_SIZE_GUARD`).
2.  **If the line is absent**: Skip Phase 3 entirely. `auto` mode already generated file/cluster summaries; `never` mode intentionally skipped.
3.  **If the line is present**: Parse the `pending_files=X pending_clusters=Y` fields, then use `AskUserQuestion` in the language specified by the loaded `language.md` directive:
    - Question (zh-CN): "检测到 {X} 个文件 + {Y} 个集群尚未生成层级摘要。立即生成？（消耗 LLM tokens）"
    - Question (en): "{X} file(s) and {Y} cluster(s) lack hierarchical summaries. Generate now? (consumes LLM tokens)"
    - Options:
        - "Yes — generate now (推荐)" — proceed to step 4.
        - "Skip this session" — proceed to step 5.
        - "Disable hierarchical summaries permanently" — proceed to step 6.
4.  **If user chose "Yes"**: Execute the bootstrap-only entry, overriding the mode to `auto`:
    ```bash
    python "~/.claude/skills/remy-index/run.py" --bootstrap-only --mode auto
    ```
    Wait for completion. The output again contains a `BOOTSTRAP_RESULT` line confirming file_done / cluster_done counts.
5.  **If user chose "Skip"**: Output a single notice: "Skipped. Run `/remy-index` again or `remy-cc summary-rebuild` later to retry."
6.  **If user chose "Disable permanently"**: Output the command for the user to apply manually: "Set `REMY_SUMMARY_BOOTSTRAP_MODE=never` through `remy-cc config --path <project>`."

### Phase 4: Injection Strategy
1.  Determine the injection policy from the effective Remy configuration key `REMY_LOGIC_INDEX_AUTO_INJECT`.
    - Default is `ALWAYS` if not set.

2.  **Branching Logic**:

    - **Case A: Policy == ALWAYS**
        - Execute Injection immediately:
          ```bash
          REMY_LOGIC_INDEX_AUTO_INJECT=ALWAYS python "~/.claude/hooks/doc_manager/injector.py"
          ```

    - **Case B: Policy == ASK**
        - Use `AskUserQuestion` to prompt: "Logic Index generated. Inject into CLAUDE.md context?"
        - Options: ["Yes (Inject)", "No (Skip)"]
        - **If Yes**:
          ```bash
          REMY_LOGIC_INDEX_AUTO_INJECT=ALWAYS python "~/.claude/hooks/doc_manager/injector.py"
          ```
        - **If No**:
          - Output: "Skipping injection."

    - **Case C: Policy == NEVER**
        - Output: "Skipping injection (Policy: NEVER)."

## Configuration

`run.py` reads the independent Remy configuration files through the shared
registry. User values live in `~/.claude/remy-config.json`; project overrides
live in `<project>/.claude/remy-config.json`. Same-named process environment
variables override files for the current process tree.

- `REMY_LLM_API_KEY`: API key for an OpenAI-compatible service.
- `REMY_LLM_MODEL`: model name (default: `deepseek-v4-flash`).

From MyPeacefulValentine/Remy-CC, MIT.

More from MyPeacefulValentine/Remy-CC