codectx/CONTEXT.md at main · hey-granth/codectx

ARCHITECTURE

codectx processes repositories through a structured analysis pipeline that ranks code by importance, compresses it intelligently, and emits a structured markdown document optimized for AI systems.

(Architecture truncated. See ARCHITECTURE.md for details.)

ENTRY_POINTS

`src/codectx/cli.py`

"""codectx CLI — typer entrypoint wiring the full pipeline."""

from __future__ import annotations

import logging
import sys
import time
from dataclasses import dataclass
from pathlib import Path

import typer
from rich.console import Console
from rich.panel import Panel
from rich.progress import Progress, SpinnerColumn, TextColumn

from codectx import __version__
from codectx.config.defaults import CACHE_DIR_NAME

app = typer.Typer(
    name="codectx",
    help="Codebase context compiler for AI agents.",
    no_args_is_help=True,
    add_completion=False,
)
console = Console(stderr=True)


@app.command()
def analyze(
    root: Path = typer.Argument(  # noqa: B008
        ".",
        help="Repository root directory to analyze.",
        exists=True,
        file_okay=False,
        resolve_path=True,
    ),
    tokens: int = typer.Option(  # noqa: B008
        None,
        "--tokens",
        "-t",
        help="Token budget (default: 120000).",
    ),
    output: Path = typer.Option(  # noqa: B008
        None,
        "--output",
        "-o",
        help="Output file path (default: CONTEXT.md).",
    ),
    since: str | None = typer.Option(  # noqa: B008
        None,
        "--since",
        help="Include recent changes since this date (e.g. '7 days ago').",
    ),
    verbose: bool = typer.Option(  # noqa: B008
        False,
        "--verbose",
        "-v",
        help="Enable verbose logging.",
    ),
    no_git: bool = typer.Option(  # noqa: B008
        False,
        "--no-git",
        help="Skip git metadata collection.",
    ),
    query: str | None = typer.Option(  # noqa: B008
        None,
        "--query",
        "-q",
        help="Semantic query to rank files by relevance (requires codectx[semantic]).",
    ),
    task: str = typer.Option(  # noqa: B008
        "default",
        "--task",
        help="Task profile for context generation (debug, feature, architecture, default).",
    ),
    layers: bool = typer.Option(  # noqa: B008
        False,
        "--layers",
        help="Generate layered context output.",
    ),
    extra_roots: list[Path] | None = typer.Option(  # noqa: B008
        None,
        "--extra-root",
        help="Additional root directories for multi-root analysis.",
    ),
) -> None:
    """Analyze a codebase and generate CONTEXT.md."""
    _setup_logging(verbose)
    start_time = time.perf_counter()

    from codectx.config.loader import load_config

    # Build roots list: primary root + any extra roots
    roots_list: list[Path] | None = None
    if extra_roots:
        roots_list = [root] + list(extra_roots)

    config = load_config(
        root,
        token_budget=tokens,
        output_file=str(output) if output else None,
        since=since,
        verbose=verbose,
        no_git=no_git,
        query=query or "",
        task=task,
        layers=layers,
        roots=roots_list,
    )

    metrics = _run_pipeline(config)
    elapsed = time.perf_counter() - start_time

    ratio = metrics.original_tokens / metrics.context_tokens if metrics.context_tokens > 0 else 0

    console.print(
        Panel(
            f"[bold green]✓[/] Context written to [bold]{metrics.output_path}[/]\n\n"
            f"[bold]Files scanned:[/] {metrics.files_scanned:,}\n"
            f"[bold]Source tokens (excl. tests/docs):[/] {metrics.original_tokens:,}\n"
            f"[bold]Context tokens:[/] {metrics.context_tokens:,}\n"
            f"[bold]Compression ratio:[/] {ratio:.1f}x\n"
            f"[bold]Analysis time:[/] {elapsed:.1f}s",
            title="codectx",
            border_style="green",
        )
    )


@app.command()
def benchmark(
    root: Path = typer.Argument(  # noqa: B008
        ".",
        help="Repository root directory.",
        exists=True,
        file_okay=False,
        resolve_path=True,
    ),
    tokens: int = typer.Option(None, "--tokens", "-t"),  # noqa: B008
    verbose: bool = typer.Option(False, "--verbose", "-v"),  # noqa: B008
    no_git: bool = typer.Option(False, "--no-git"),  # noqa: B008
) -> None:
    """Run analysis with detailed timing and stats."""
    _setup_logging(verbose)

    from codectx.config.loader import load_config

    config = load_config(
        root,
        token_budget=tokens,
        verbose=verbose,
        no_git=no_git,
    )

    console.print("[bold]Running benchmark...[/]\n")

    timings: dict[str, float] = {}

    # Walk
    t0 = time.perf_counter()
    from codectx.walker import walk

    files = walk(config.root, config.extra_ignore)
    timings["walk"] = time.perf_counter() - t0

    # Parse
    t0 = time.perf_counter()
    from codectx.parser.treesitter import parse_files

    parse_results = parse_files(files)
    timings["parse"] = time.perf_counter() - t0

    # Graph
    t0 = time.perf_counter()
    from codectx.graph.builder import build_dependency_graph

    dep_graph = build_dependency_graph(parse_results, config.root)
    timings["graph"] = time.perf_counter() - t0

    # Rank
    t0 = time.perf_counter()
    from codectx.ranker.git_meta import collect_git_metadata
    from codectx.ranker.scorer import score_files

    git_meta = collect_git_metadata(files, config.root, config.no_git)
    scores = score_files(files, dep_graph, git_meta)
    timings["rank"] = time.perf_counter() - t0

    # Compress
    t0 = time.perf_counter()
    from codectx.compressor.budget import TokenBudget
    from codectx.compressor.tiered import compress_files

    budget = TokenBudget(config.token_budget)
    compressed = compress_files(parse_results, scores, budget, config.root)
    timings["compress"] = time.perf_counter() - t0

    total = sum(timings.values())

    console.print(
        Panel(
            "\n".join(
                [
                    f"[bold]Files discovered:[/] {len(files)}",
                    f"[bold]Files parsed:[/] {len(parse_results)}",
                    f"[bold]Graph nodes:[/] {dep_graph.node_count}",
                    f"[bold]Graph edges:[/] {dep_graph.edge_count}",
                    f"[bold]Compressed files:[/] {len(compressed)}",
                    f"[bold]Tokens used:[/] {budget.used:,} / {budget.total:,}",
                    "",
                    *[f"  {k:>10}: {v:.3f}s" for k, v in timings.items()],
                    f"  {'total':>10}: {total:.3f}s",
                ]
            ),
            title="Benchmark Results",
            border_style="cyan",
        )
    )


@app.command()
def watch(
    root: Path = typer.Argument(  # noqa: B008
        ".",
        help="Repository root directory.",
        exists=True,
        file_okay=False,
        resolve_path=True,
    ),
    tokens: int = typer.Option(None, "--tokens", "-t"),  # noqa: B008
    output: Path = typer.Option(None, "--output", "-o"),  # noqa: B008
    verbose: bool = typer.Option(False, "--verbose", "-v"),  # noqa: B008
    no_git: bool = typer.Option(False, "--no-git"),  # noqa: B008
) -> None:
    """Watch for file changes and regenerate CONTEXT.md."""
    _setup_logging(verbose)

    from codectx.config.loader import load_config

    config = load_config(
        root,
        token_budget=tokens,
        output_file=str(output) if output else None,
        verbose=verbose,
        no_git=no_git,
        watch=True,
    )

    console.print(f"[bold]Watching[/] {config.root} for changes...")
    console.print("Press Ctrl+C to stop.\n")

    # Initial run
    _run_pipeline(config)
    console.print("[green]Initial context generated.[/]\n")

    try:
        from watchfiles import watch as watchfiles_watch

        for changes in watchfiles_watch(str(config.root)):
            changed_paths = [Path(c[1]) for c in changes]
            console.print(f"[yellow]Changes detected:[/] {len(changed_paths)} file(s)")
            try:
                _run_pipeline(config)
                console.print("[green]Context regenerated.[/]\n")
            except Exception as exc:
                console.print(f"[red]Error during regeneration: {exc}[/]\n")
    except KeyboardInterrupt:
        console.print("\n[bold]Watch stopped.[/]")


@app.command()
def search(
    query: str = typer.Argument(  # noqa: B008
        ...,
        help="Semantic search query.",
    ),
    root: Path = typer.Option(  # noqa: B008
        ".",
        "--root",
        "-r",
        help="Repository root directory.",
        exists=True,
        file_okay=False,
        resolve_path=True,
    ),
    limit: int = typer.Option(  # noqa: B008
        10,
        "--limit",
        "-l",
        help="Number of results to return.",
    ),
    verbose: bool = typer.Option(  # noqa: B008
        False,
        "--verbose",
        "-v",
        help="Enable verbose logging.",
    ),
) -> None:
    """Search the codebase semantically."""
    _setup_logging(verbose)

... (truncated: entry point exceeds 300 lines)

SYMBOL_INDEX

src/codectx/parser/base.py

class Symbol
class ParseResult
make_plaintext_result()

src/codectx/cli.py

analyze()
benchmark()
watch()
search()
cache_export()
cache_import()
main()
class PipelineMetrics
_run_pipeline()
_setup_logging()

src/codectx/output/formatter.py

_root_label()
format_context()
write_context_file()
write_layer_files()
_section_header()
_auto_architecture()
_render_mermaid_graph()

src/codectx/graph/builder.py

class DepGraph
- add_file()
- add_edge()
- fan_in()
- fan_out()
- entry_points()
- graph_distance()
- entry_distances()
- detect_call_paths()
build_dependency_graph()

src/codectx/parser/treesitter.py

_parse_scm_patterns()
class QuerySpec
_load_query_spec()
_get_query_spec()
parse_files()
parse_file()
_parse_single_worker()
_log_parse_health()
_extract()
_fallback_parse()
_regex_imports()
_regex_docstrings()
_extract_imports()
_extract_symbols()
_extract_module_docstrings()
_python_func_symbol()
_python_class_symbol()
_js_func_symbol()
_js_class_symbol()
_maybe_js_arrow()
_go_func_symbol()
_generic_symbol()
_walk_tree()
_node_text()
_find_child()
_extract_first_docstring()
_read_source()

src/codectx/ranker/scorer.py

score_files()
_min_max_normalize()

src/codectx/compressor/tiered.py

class CompressedFile
_is_non_source()
assign_tiers()
compress_files()
_tier1_content()
_extract_internal_imports()
_structured_summary_content()
_tier2_content()
_tier3_content()
_one_line_summary()

src/codectx/walker.py

walk()
_collect()
_is_binary()
walk_multi()
find_root()

src/codectx/ranker/git_meta.py

class GitFileInfo
collect_git_metadata()
_collect_from_git()
_filesystem_fallback()
collect_recent_changes()
_parse_since()
_load_pygit2()

src/codectx/ranker/semantic.py

is_available()
semantic_score()

src/codectx/cache.py

class Cache
- __init__()
- _load()
- save()
- get_parse_result()
- put_parse_result()
- get_token_count()
- put_token_count()
- invalidate()
- export_cache()
file_hash()
_decode_children()
_coerce_int()

src/codectx/config/loader.py

class Config
load_config()
_resolve()
_resolve_bool()
_resolve_str()
_resolve_optional_str()
_resolve_int()

src/codectx/parser/languages.py

class LanguageEntry
class TreeSitterLanguageLoadError
get_language()
get_language_for_path()
get_ts_language_object()
_coerce_language()
load_typescript_language()
supported_extensions()

src/codectx/graph/resolver.py

resolve_import()
resolve_import_multi_root()
_resolve_python()
_resolve_js_ts()
_resolve_go()
_resolve_rust()
_resolve_java()
_resolve_c_cpp()
_resolve_ruby()

src/codectx/output/sections.py

class Section

src/codectx/compressor/budget.py

_get_encoder()
count_tokens()
class TokenBudget
- __init__()
- consume()
- consume_partial()

src/codectx/compressor/summarizer.py

is_available()
summarize_file()
summarize_files_batch()
_summarize_openai()
_summarize_anthropic()

main.py

main()

IMPORTANT_CALL_PATHS

main.main()

CORE_MODULES

`src/codectx/parser/base.py`

Purpose: Core data structures for the parser module.

Types:

ParseResult - Result of parsing a single source file.
Symbol - A top-level symbol extracted from a source file.

Functions:

def make_plaintext_result(path: Path, source: str) -> ParseResult
- Create a minimal ParseResult for unsupported language files.

`src/codectx/config/defaults.py`

Purpose: Default configuration values and constants for codectx.

`src/codectx/output/formatter.py`

Purpose: Structured markdown formatter — emits CONTEXT.md. Depends on: compressor.tiered, config.defaults, graph.builder, output.sections, +1 more

Functions:

def _auto_architecture(compressed: list[CompressedFile], root: Path) -> str
def _render_mermaid_graph( dep_graph: DepGraph, root: Path, compressed: list[CompressedFile], ) -> str
def _root_label(file_path: Path, roots: list[Path] | None) -> str
def _section_header(title: str) -> str

`src/codectx/graph/builder.py`

Purpose: Dependency graph construction using rustworkx. Depends on: config.defaults, graph.resolver, parser.base

Types:

DepGraph - Dependency graph with file-level nodes and import edges. methods: add_edge, add_file, detect_call_paths, entry_distances, entry_points, fan_in (+2 more)

Functions:

def build_dependency_graph( parse_results: dict[Path, ParseResult], root: Path, ) -> DepGraph
- Build a dependency graph from parse results.

`src/codectx/parser/treesitter.py`

Purpose: Tree-sitter AST extraction — parallel parsing of source files. Depends on: config.defaults, parser.base, parser.languages

Types:

QuerySpec - Parsed query specification from a .scm file.

Functions:

def _extract(path: Path, source: str, entry: LanguageEntry) -> ParseResult
def _extract_first_docstring(body_node: Any, source: str) -> str
def _extract_imports(node: Any, language: str, source: str) -> list[str]
def _extract_module_docstrings(node: Any, language: str, source: str) -> list[str]

`src/codectx/ranker/scorer.py`

Purpose: Composite file scoring — ranks files by importance. Depends on: config.defaults, graph.builder, parser.base, ranker.git_meta

Functions:

def _min_max_normalize(values: dict[Path, float]) -> dict[Path, float]
- Min-max normalize values to [0, 1]. Returns 0 for all if constant.
def score_files(files: list[Path], dep_graph: DepGraph, git_meta: dict[Path, GitFileInfo], ...) -> dict[Path, float]
- Score each file 0.0–1.0 using a weighted composite.

`src/codectx/compressor/tiered.py`

Purpose: Tiered compression — assigns tiers and enforces token budget. Depends on: compressor.budget, compressor.summarizer, config.defaults, parser.base

Types:

CompressedFile - A file compressed to its assigned tier.

Functions:

def _extract_internal_imports(imports: tuple[str, ...], root: Path, source_path: Path) -> list[str]
def _is_non_source(path: Path, root: Path) -> bool
def _one_line_summary(pr: ParseResult) -> str
def _structured_summary_content(pr: ParseResult, path: Path, root: Path) -> str

`src/codectx/walker.py`

Purpose: File-system walker — discovers files, applies ignore specs, filters binaries. Depends on: config.defaults, ignore

Functions:

def _collect( current: Path, root: Path, spec: pathspec.PathSpec, out: list[Path], ) -> None
def _is_binary(path: Path) -> bool
def find_root(file_path: Path, roots: list[Path]) -> Path | None
def walk( root: Path, extra_ignore: tuple[str, ...] = (), output_file: Path | None = None, ) -> list[Path]
def walk_multi(roots: list[Path], ...), output_file: Path | None = None, ) -> dict[Path, list[Path]]

`src/codectx/ranker/git_meta.py`

Purpose: Git metadata extraction via pygit2.

Types:

GitFileInfo - Git metadata for a single file.

Functions:

def _collect_from_git(repo: Any, pygit2_mod: Any, files: list[Path], root: Path, ...) -> dict[Path, GitFileInfo]
def _filesystem_fallback(files: list[Path]) -> dict[Path, GitFileInfo]
def _load_pygit2() -> Any | None
def _parse_since(since: str) -> float | None
def collect_git_metadata(files: list[Path], root: Path, no_git: bool = False, ...) -> dict[Path, GitFileInfo]
def collect_recent_changes(root: Path, since: str | None, no_git: bool = False) -> str

`pyproject.toml`

Purpose: Implements pyproject.

`src/codectx/ranker/semantic.py`

Purpose: Semantic search ranking using lancedb and sentence-transformers. Depends on: parser.base

Functions:

def is_available() -> bool
- Check if semantic search dependencies are available.
def semantic_score(query: str, files: list[Path], parse_results: dict[Path, ParseResult], ...) -> dict[Path, float]
- Return semantic relevance score 0.0–1.0 per file for the given query.

`src/codectx/cache.py`

Purpose: File-level caching for parse results, token counts, and git metadata. Depends on: config.defaults, parser.base

Types:

Cache - JSON-based file cache in .codectx_cache/. methods: __init__, export_cache, get_parse_result, get_token_count, invalidate, put_parse_result (+2 more)

Functions:

def _coerce_int(value: object) -> int | None
def _decode_children(children: list[Any] | tuple[Any, ...]) -> tuple[Symbol, ...]
def file_hash(path: Path) -> str
- Compute a fast hash of file contents.

SUPPORTING_MODULES

`README.md`

215 lines, 0 imports

`src/codectx/config/loader.py`

Configuration loader — reads .codectx.toml or pyproject.toml [tool.codectx].

class Config
    """Resolved configuration for a codectx run."""

def load_config(root: Path, **cli_overrides: object) -> Config
    """Load config from .codectx.toml → pyproject.toml [tool.codectx] → defaults.

    CLI overrides take highest precedence."""

def _resolve(
    key: str,
    cli: dict[str, object],
    file_cfg: dict[str, object],
    default: object,
) -> object
    """Resolve a config key with precedence: CLI > file > default."""

def _resolve_bool(
    key: str,
    cli: dict[str, object],
    file_cfg: dict[str, object],
    default: bool,
) -> bool

def _resolve_str(
    key: str,
    cli: dict[str, object],
    file_cfg: dict[str, object],
    default: str,
) -> str

def _resolve_optional_str(
    key: str,
    cli: dict[str, object],
    file_cfg: dict[str, object],
    default: str | None,
) -> str | None

def _resolve_int(
    key: str,
    cli: dict[str, object],
    file_cfg: dict[str, object],
    default: int,
) -> int

`src/codectx/parser/languages.py`

Extension → language mapping for tree-sitter parsers.

class LanguageEntry
    """A supported language with its tree-sitter module reference."""

class TreeSitterLanguageLoadError(RuntimeError)
    """Raised when a tree-sitter language cannot be resolved safely."""

def get_language(ext: str) -> LanguageEntry | None
    """Return the LanguageEntry for a file extension, or None if unsupported."""

def get_language_for_path(path: Any) -> LanguageEntry | None
    """Return the LanguageEntry for a file path (uses suffix)."""

def get_ts_language_object(entry: LanguageEntry) -> Any
    """Dynamically import and return the tree-sitter Language object.

    Uses the modern per-package tree-sitter bindings (tree-sitter-python, etc.)."""

def _coerce_language(value: Any) -> tree_sitter.Language
    """Normalize any supported language payload into a Language object."""

def load_typescript_language(language_fn: str = "language_typescript") -> tree_sitter.Language
    """Load TypeScript grammar across tree_sitter_typescript API variants.

    Supported exports across known package versions include:
    - callable factories: language(), get_language(), language_typescript(), language_tsx()
    - constants: LANGUAGE, LANGUAGE_TYPESCRIPT, LANGUAGE_TSX
    - manual binding fallback via tree_sitter.Language(<shared-library>, <name>)"""

def supported_extensions() -> frozenset[str]
    """Return all file extensions supported for tree-sitter parsing."""

`src/codectx/graph/resolver.py`

Per-language import string → file path resolution.

def resolve_import(
    import_text: str,
    language: str,
    source_file: Path,
    root: Path,
    all_files: frozenset[str],
) -> list[Path]
    """Resolve an import statement to file paths within the repository.

    Args:
        import_text: Raw import string from the AST.
        language: Language name (e.g. "python").
        source_file: Absolute path of the file containing the import.
        root: Repository root.
        all_files: Set of all known file paths (POSIX, relative to root).

    Returns:
        List of resolved file paths (may be empty if unresolvable)."""

def resolve_import_multi_root(
    import_text: str,
    language: str,
    source_file: Path,
    roots: list[Path],
    all_files_by_root: dict[Path, frozenset[str]],
) -> list[Path]
    """Resolve an import trying the source file's root first, then others.

    Args:
        import_text: Raw import string from the AST.
        language: Language name.
        source_file: Absolute path of the file containing the import.
        roots: All root directories.
        all_files_by_root: Map of root → set of relative file paths.

    Returns:
        List of resolved file paths."""

def _resolve_python(
    import_text: str,
    source_file: Path,
    root: Path,
    all_files: frozenset[str],
) -> list[Path]

def _resolve_js_ts(
    import_text: str,
    source_file: Path,
    root: Path,
    all_files: frozenset[str],
) -> list[Path]

def _resolve_go(import_text: str, root: Path, all_files: frozenset[str]) -> list[Path]

def _resolve_rust(
    import_text: str,
    source_file: Path,
    root: Path,
    all_files: frozenset[str],
) -> list[Path]

def _resolve_java(import_text: str, root: Path, all_files: frozenset[str]) -> list[Path]

def _resolve_c_cpp(
    import_text: str,
    source_file: Path,
    root: Path,
    all_files: frozenset[str],
) -> list[Path]

def _resolve_ruby(
    import_text: str,
    source_file: Path,
    root: Path,
    all_files: frozenset[str],
) -> list[Path]

`src/codectx/output/sections.py`

Section constants for CONTEXT.md output.

class Section
    """A named section in the output file."""

`src/codectx/compressor/budget.py`

Token counting and budget tracking via tiktoken.

def _get_encoder() -> tiktoken.Encoding

def count_tokens(text: str) -> int
    """Count the number of tokens in *text*."""

class TokenBudget
    """Tracks remaining token budget during context assembly."""

`src/codectx/init.py`

codectx — Codebase context compiler for AI agents.

4 lines, 0 imports

`.gitignore`

44 lines, 0 imports

`PLAN.md`

145 lines, 0 imports

`src/codectx/compressor/summarizer.py`

LLM-based file summarization for Tier 3 compression.

This module is an optional dependency — all LLM imports are guarded. Install with: pip install codectx[llm]

def is_available() -> bool
    """Check if any LLM provider is available."""

def summarize_file(result: ParseResult, provider: str = "openai", model: str = "") -> str
    """Return one-sentence summary of the file's purpose.

    Args:
        result: ParseResult for the file.
        provider: LLM provider ('openai' or 'anthropic').
        model: Model name (defaults to provider-specific default).

    Returns:
        One-sentence summary string.

    Raises:
        ImportError: If the required provider is not installed.
        RuntimeError: If the summarization call fails."""

def summarize_files_batch(
    results: list[ParseResult],
    provider: str = "openai",
    model: str = "",
    max_workers: int = 4,
) -> dict[Path, str]
    """Summarize multiple files concurrently.

    Args:
        results: List of ParseResult objects to summarize.
        provider: LLM provider name.
        model: Model name.
        max_workers: Max concurrent summarization threads.

    Returns:
        Dict mapping file path to summary string."""

def _summarize_openai(prompt: str, model: str) -> str
    """Call OpenAI API for summarization."""

def _summarize_anthropic(prompt: str, model: str) -> str
    """Call Anthropic API for summarization."""

`.dockerignore`

27 lines, 0 imports

`Dockerfile`

48 lines, 0 imports

`docker-compose.yml`

11 lines, 0 imports

`main.py`

def main()

DEPENDENCY_GRAPH

graph LR
    f0["src/codectx/parser/base.py"]
    f1["src/codectx/cli.py"]
    f2["src/codectx/output/formatter.py"]
    f3["src/codectx/graph/builder.py"]
    f4["src/codectx/parser/treesitter.py"]
    f5["src/codectx/ranker/scorer.py"]
    f6["src/codectx/compressor/tiered.py"]
    f7["src/codectx/walker.py"]
    f8["src/codectx/ranker/git_meta.py"]
    f9["pyproject.toml"]
    f10["src/codectx/ranker/semantic.py"]
    f11["src/codectx/cache.py"]
    f12["src/codectx/parser/languages.py"]
    f13["src/codectx/graph/resolver.py"]
    f14["src/codectx/output/sections.py"]
    f15["src/codectx/compressor/budget.py"]
    f16["src/codectx/__init__.py"]
    f17[".gitignore"]
    f18["src/codectx/compressor/summarizer.py"]
    f19[".dockerignore"]
    f20["Dockerfile"]
    f21["docker-compose.yml"]
    f22["main.py"]
    f23["src/codectx/ignore.py"]
    f24["src/codectx/safety.py"]
    f1 --> f24
    f1 --> f2
    f1 --> f0
    f1 --> f11
    f1 --> f10
    f1 --> f6
    f1 --> f15
    f1 --> f5
    f1 --> f8
    f1 --> f3
    f1 --> f4
    f1 --> f7
    f1 --> f16
    f2 --> f0
    f2 --> f14
    f2 --> f3
    f2 --> f6
    f3 --> f0
    f3 --> f13
    f4 --> f12
    f4 --> f0
    f5 --> f8
    f5 --> f0
    f5 --> f3
    f6 --> f18
    f6 --> f0
    f6 --> f15
    f7 --> f23
    f10 --> f0
    f11 --> f0
    f18 --> f0

RANKED_FILES

File	Score	Tier	Tokens
`src/codectx/parser/base.py`	0.653	structured summary	91
`src/codectx/cli.py`	0.653	full source	2189
`src/codectx/config/defaults.py`	0.624	structured summary	25
`src/codectx/output/formatter.py`	0.583	structured summary	150
`src/codectx/graph/builder.py`	0.574	structured summary	134
`src/codectx/parser/treesitter.py`	0.512	structured summary	161
`src/codectx/ranker/scorer.py`	0.503	structured summary	150
`src/codectx/compressor/tiered.py`	0.438	structured summary	164
`src/codectx/walker.py`	0.423	structured summary	186
`src/codectx/ranker/git_meta.py`	0.414	structured summary	194
`pyproject.toml`	0.390	structured summary	14
`src/codectx/ranker/semantic.py`	0.359	structured summary	115
`src/codectx/cache.py`	0.357	structured summary	160
`README.md`	0.287	signatures	13
`src/codectx/config/loader.py`	0.282	signatures	300
`src/codectx/parser/languages.py`	0.260	signatures	320
`tests/unit/test_treesitter.py`	0.237	one-liner	18
`tests/unit/test_semantic.py`	0.218	one-liner	17
`src/codectx/graph/resolver.py`	0.218	signatures	537
`src/codectx/output/sections.py`	0.216	signatures	38
`tests/test_integration.py`	0.194	one-liner	20
`tests/test_parser.py`	0.183	one-liner	15
`src/codectx/compressor/budget.py`	0.175	signatures	74
`tests/unit/test_cache_export.py`	0.158	one-liner	17
`docs/astro.config.mjs`	0.153	one-liner	18
`tests/test_scorer.py`	0.148	one-liner	17
`tests/test_walker.py`	0.148	one-liner	15
`tests/unit/test_git_meta.py`	0.148	one-liner	16
`tests/unit/test_formatter_coverage.py`	0.141	one-liner	15
`tests/unit/test_formatter_sections.py`	0.141	one-liner	19
`src/codectx/__init__.py`	0.137	signatures	33
`.gitignore`	0.134	signatures	13
`PLAN.md`	0.115	signatures	13
`tests/test_compressor.py`	0.114	one-liner	18
`tests/unit/test_summarizer.py`	0.114	one-liner	19
`src/codectx/compressor/summarizer.py`	0.110	signatures	348
`tests/unit/test_resolver.py`	0.106	one-liner	14
`tests/unit/test_multi_root.py`	0.104	one-liner	16
`.dockerignore`	0.100	signatures	13
`Dockerfile`	0.100	signatures	14

PERIPHERY

tests/unit/test_treesitter.py — Tests for multi-language treesitter parsing.
tests/unit/test_semantic.py — Tests for semantic search ranking module.
tests/test_integration.py — Integration test — runs codectx pipeline end-to-end.
tests/test_parser.py — Tests for tree-sitter parsing.
tests/unit/test_cache_export.py — Tests for CI cache export/import.
docs/astro.config.mjs — 2 imports, 75 lines
tests/test_scorer.py — Tests for the composite file scorer.
tests/test_walker.py — Tests for the file walker.
tests/unit/test_git_meta.py — Tests for git metadata collection.
tests/unit/test_formatter_coverage.py — Tests for output formatting.
tests/unit/test_formatter_sections.py — Tests for deterministic formatter section ordering and presence.
tests/test_compressor.py — Tests for tiered compression and token budget.
tests/unit/test_summarizer.py — Tests for LLM summarizer module.
tests/unit/test_resolver.py — Tests for import resolution.
tests/unit/test_multi_root.py — Tests for multi-root support.
docs/src/content/docs/guides/docker.md — 74 lines
tests/unit/test_call_paths.py — Tests for call path detection and formatting.
tests/unit/test_safety.py — Tests for safety checks in pipeline flow.
ARCHITECTURE.md — 252 lines
DECISIONS.md — 262 lines
tests/unit/test_cycles.py — Tests for cyclic dependency detection.
src/codectx/ignore.py — Ignore-spec handling — layers ALWAYS_IGNORE, .gitignore, .ctxignore.
tests/unit/test_cli.py — Tests for CLI commands.
tests/unit/test_cache_wiring.py — Tests for cache wiring into the analyze pipeline.
src/codectx/safety.py — Sensitive-file detection and user confirmation.
docs/src/content/docs/advanced/dependency-graph.md — 23 lines
docs/src/content/docs/advanced/ranking-system.md — 41 lines
docs/src/content/docs/advanced/token-compression.md — 27 lines
docs/src/content/docs/comparison.md — 31 lines
docs/src/content/docs/getting-started/basic-usage.md — 63 lines
docs/src/content/docs/getting-started/installation.md — 62 lines
docs/src/content/docs/getting-started/quick-start.mdx — 44 lines
docs/src/content/docs/guides/configuration.md — 53 lines
docs/src/content/docs/introduction/what-is-codectx.md — 22 lines
docs/src/content/docs/reference/architecture-overview.md — 33 lines
docs/src/content/docs/reference/cli-reference.md — 116 lines
tests/unit/test_queries.py — Tests for .scm query file loading and data-driven extraction.
tests/unit/test_semantic_mock.py — Mock tests for semantic logic.
tests/test_ignore.py — Tests for ignore-spec handling.
docs/package.json — 26 lines
docs/src/content.config.ts — 3 imports, 7 lines
docs/build_output.txt — 382 lines
docs/src/content/docs/community/contributing.md — 52 lines
docs/src/content/docs/community/faq.md — 23 lines
docs/src/content/docs/guides/best-practices.md — 34 lines
docs/src/content/docs/guides/using-context-effectively.md — 34 lines
docs/src/content/docs/index.mdx — 32 lines
docs/src/content/docs/introduction/why-it-exists.md — 20 lines
docs/src/env.d.ts — 3 lines
docs/src/styles/custom.css — 19 lines
docs/tsconfig.json — 10 lines
src/codectx/parser/queries/go.scm — 7 lines
src/codectx/parser/queries/java.scm — 5 lines
src/codectx/parser/queries/javascript.scm — 8 lines
src/codectx/parser/queries/python.scm — 7 lines
src/codectx/parser/queries/rust.scm — 8 lines
src/codectx/parser/queries/typescript.scm — 8 lines
tests/unit/__init__.py — 0 lines
src/codectx/compressor/__init__.py — 0 lines
src/codectx/config/__init__.py — 0 lines
src/codectx/graph/__init__.py — 0 lines
src/codectx/output/__init__.py — 0 lines
src/codectx/parser/__init__.py — 0 lines
src/codectx/ranker/__init__.py — 0 lines
tests/__init__.py — 0 lines
.python-version — 2 lines

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ARCHITECTURE

ENTRY_POINTS

`src/codectx/cli.py`

SYMBOL_INDEX

IMPORTANT_CALL_PATHS

CORE_MODULES

`src/codectx/parser/base.py`

`src/codectx/config/defaults.py`

`src/codectx/output/formatter.py`

`src/codectx/graph/builder.py`

`src/codectx/parser/treesitter.py`

`src/codectx/ranker/scorer.py`

`src/codectx/compressor/tiered.py`

`src/codectx/walker.py`

`src/codectx/ranker/git_meta.py`

`pyproject.toml`

`src/codectx/ranker/semantic.py`

`src/codectx/cache.py`

SUPPORTING_MODULES

`README.md`

`src/codectx/config/loader.py`

`src/codectx/parser/languages.py`

`src/codectx/graph/resolver.py`

`src/codectx/output/sections.py`

`src/codectx/compressor/budget.py`

`src/codectx/init.py`

`.gitignore`

`PLAN.md`

`src/codectx/compressor/summarizer.py`

`.dockerignore`

`Dockerfile`

`docker-compose.yml`

`main.py`

DEPENDENCY_GRAPH

RANKED_FILES

PERIPHERY

FilesExpand file tree

CONTEXT.md

Latest commit

History

CONTEXT.md

File metadata and controls

ARCHITECTURE

ENTRY_POINTS

src/codectx/cli.py

SYMBOL_INDEX

IMPORTANT_CALL_PATHS

CORE_MODULES

src/codectx/parser/base.py

src/codectx/config/defaults.py

src/codectx/output/formatter.py

src/codectx/graph/builder.py

src/codectx/parser/treesitter.py

src/codectx/ranker/scorer.py

src/codectx/compressor/tiered.py

src/codectx/walker.py

src/codectx/ranker/git_meta.py

pyproject.toml

src/codectx/ranker/semantic.py

src/codectx/cache.py

SUPPORTING_MODULES

README.md

src/codectx/config/loader.py

src/codectx/parser/languages.py

src/codectx/graph/resolver.py

src/codectx/output/sections.py

src/codectx/compressor/budget.py

src/codectx/__init__.py

.gitignore

PLAN.md

src/codectx/compressor/summarizer.py

.dockerignore

Dockerfile

docker-compose.yml

main.py

DEPENDENCY_GRAPH

RANKED_FILES

PERIPHERY

`src/codectx/cli.py`

`src/codectx/parser/base.py`

`src/codectx/config/defaults.py`

`src/codectx/output/formatter.py`

`src/codectx/graph/builder.py`

`src/codectx/parser/treesitter.py`

`src/codectx/ranker/scorer.py`

`src/codectx/compressor/tiered.py`

`src/codectx/walker.py`

`src/codectx/ranker/git_meta.py`

`pyproject.toml`

`src/codectx/ranker/semantic.py`

`src/codectx/cache.py`

`README.md`

`src/codectx/config/loader.py`

`src/codectx/parser/languages.py`

`src/codectx/graph/resolver.py`

`src/codectx/output/sections.py`

`src/codectx/compressor/budget.py`

`src/codectx/init.py`

`.gitignore`

`PLAN.md`

`src/codectx/compressor/summarizer.py`

`.dockerignore`

`Dockerfile`

`docker-compose.yml`

`main.py`