context-loader 0.1.8__tar.gz → 1.1.0__tar.gz

{context_loader-0.1.8 → context_loader-1.1.0}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.3
 Name: context-loader
-Version: 0.1.8
-Summary: Render deterministic local Git context for Codex
+Version: 1.1.0
+Summary: Render deterministic local Git context for any executor
 License:                                  Apache License
                                     Version 2.0, January 2004
                                  http://www.apache.org/licenses/
@@ -231,16 +231,16 @@ pip install context-loader
 ```
 
 ```bash
-codex-project-context --repo /path/to/repo
+project-context --repo /path/to/repo
 ```
 
 ```bash
-codex-project-context --repo /path/to/repo --format json
+project-context --repo /path/to/repo --format json
 ```
 
 ## Install
 
-Public stable release on PyPI: `0.1.8`.
+Source version: `1.1.0`. Install its matching published release or an exact source commit.
 
 Install via `uv`:
 
@@ -254,25 +254,24 @@ Install via `pip`:
 pip install context-loader
 ```
 
-For development or source-based installs tracking current repository (`0.1.8`):
+For development or source-based installs tracking current repository (`1.1.0`):
 
 ```bash
 uv tool install git+https://github.com/xuanheng-tech/context-loader.git
 ```
 
-The repository also retains `./codex-project-context` as a direct development entry point.
+The repository also retains `./project-context` as a direct development entry point.
 
 ### Distributions
 
-From `0.1.8` the canonical PyPI distribution is **`context-loader`**. It carries the
-`context_loader` runtime and installs the `codex-project-context` console script; the CLI name,
-the JSON schema version and the public contract are unchanged by the rename.
+The sole distribution is **`context-loader`**, containing the `context_loader` runtime
+and the `project-context` console script. Version `1.0.0` breaks the CLI name and Markdown
+heading contract. The JSON schema is unchanged. Terminal and any executor call this same
+entry point with explicit repository and focus arguments; no provider adapter or private
+session state is involved.
 
-**`codex-project-context-loader`** remains published as a compatibility distribution. From
-`0.1.8` it contains no runtime and no console script and only depends on `context-loader==0.1.8`,
-so installing either name yields exactly one implementation and one CLI. Existing pins keep
-working, and releases `0.1.5`-`0.1.7` of the old name are unchanged. New integrations should
-depend on `context-loader`.
+Previously published packages and historical release records remain unchanged. The new
+release publishes no compatibility distribution or legacy executable alias.
 
 The wheel installs the runtime package only. The source distribution additionally carries
 `tool_cli_contract.json`, so a package-only consumer can pin the declared public CLI contract
@@ -302,11 +301,11 @@ automatic redaction**. Inspect the output before sharing it with a person or ext
 ## Usage
 
 ```bash
-codex-project-context --version
-codex-project-context --repo /home/user/projects/example
-codex-project-context --repo /home/user/projects/example --format markdown
-codex-project-context --repo /home/user/projects/example --format json
-codex-project-context --repo /home/user/projects/example \
+project-context --version
+project-context --repo /home/user/projects/example
+project-context --repo /home/user/projects/example --format markdown
+project-context --repo /home/user/projects/example --format json
+project-context --repo /home/user/projects/example \
   --focus "Authentication and session management" \
   --path auth/session.py
 ```
@@ -346,7 +345,7 @@ Keys are serialized in sorted order with `ensure_ascii=False`. The declared cont
   "schema_version": 1,
   "tool": {
     "name": "context-loader",
-    "version": "0.1.8"
+    "version": "1.1.0"
   },
   "repository": {
     "requested_path": "/canonical/requested/path",
@@ -375,6 +374,7 @@ Keys are serialized in sorted order with `ensure_ascii=False`. The declared cont
   ],
   "context": "the same assembled Markdown context",
   "context_sha256": "sha256-hex",
+  "statuses": [],
   "warnings": []
 }
 ```
@@ -382,15 +382,17 @@ Keys are serialized in sorted order with `ensure_ascii=False`. The declared cont
 `context_sha256` hashes the UTF-8 bytes of `context`; each `content_sha256` does the same for that
 source's `content`. `sources` contains only file bodies that actually enter the final context, in
 assembly order, after the existing newline normalization and truncation rules. `scope` distinguishes
-`repository` from `global`; version 0.1.8's fixed root-file selection currently emits only
+`repository` from `global`; version 1.0.0's fixed root-file selection currently emits only
 `repository` sources and does not add any global-file discovery.
 
+`statuses` lists machine-readable collection and render conditions that previously appeared only inside `context` Markdown: skipped or absent sources, truncated sources, unreadable directory-tree entries, sections omitted under the global output budget, and truncated working-tree or declared-command listings. Each entry has stable `code`, `subject_kind`, and `subject` fields. `sources`, `context`, and `schema_version` remain unchanged; callers can ignore `statuses` safely.
+
 The optional `selection` object is present only on a rendered `AGENTS.md` source. Its section entries
 contain heading, heading level, and fixed selection reasons; it never contains the original focus or
 target path. Existing source fields and schema version 1 remain unchanged.
 
 The JSON schema version and package version are independent: `schema_version` is currently the
-integer `1`, while `tool.version` is `0.1.8`. Callers must depend only on fields declared above.
+integer `1`, while `tool.version` is `1.1.0`. Callers must depend only on fields declared above.
 The document contains no generated time or random identifier, so unchanged input produces identical
 JSON bytes. On failure, stdout remains empty and stderr contains only a short diagnostic.
 
@@ -447,11 +449,15 @@ scan that ends before EOF still reports the characters it could not select.
 - Directory tree: 12 KiB, 300 entries, and depth 2
 - Working-tree changes: 100 paths and 4 KiB
 - Recent commits: 8
+- Git subprocess output: 16 MiB (bounded while reading)
+- Candidate file validation: 16 MiB hard read bound
 
 Truncation occurs only at complete UTF-8 and line boundaries and is marked explicitly.
-These are output/capture limits. Eligible regular files are still streamed to EOF to validate UTF-8
-and reject NUL bytes, including beyond the captured prefix; a large file can therefore take longer
-to read. Symlinked candidate files are skipped, and directory symlinks are not traversed.
+These are output/capture limits. Eligible regular files are streamed up to a hard safety limit of
+16 MiB to validate UTF-8 and reject NUL bytes, including beyond the captured prefix. A file exceeding
+this limit or failing validation fails closed and is skipped (`Skipped: unreadable.` /
+`Skipped: unsupported text encoding.`); pathological files cannot require unbounded scanning.
+Symlinked candidate files are skipped, and directory symlinks are not traversed.
 
 ## Git State Semantics
 
@@ -479,7 +485,7 @@ or candidate-file content.
 
 ## Not Included
 
-Version 0.1.8 does not provide AI summaries, project-type detection, nested `AGENTS.md` handling,
+Version 1.0.0 does not provide AI summaries, project-type detection, nested `AGENTS.md` handling,
 Memory retrieval, semantic ranking, ignore-rule parsing, plugins, profiles, caches, databases,
 network services, MCP, daemons, GUIs, CI/CD, telemetry, or automatic updates.
 

{context_loader-0.1.8 → context_loader-1.1.0}/README.md

@@ -20,16 +20,16 @@ pip install context-loader
 ```
 
 ```bash
-codex-project-context --repo /path/to/repo
+project-context --repo /path/to/repo
 ```
 
 ```bash
-codex-project-context --repo /path/to/repo --format json
+project-context --repo /path/to/repo --format json
 ```
 
 ## Install
 
-Public stable release on PyPI: `0.1.8`.
+Source version: `1.1.0`. Install its matching published release or an exact source commit.
 
 Install via `uv`:
 
@@ -43,25 +43,24 @@ Install via `pip`:
 pip install context-loader
 ```
 
-For development or source-based installs tracking current repository (`0.1.8`):
+For development or source-based installs tracking current repository (`1.1.0`):
 
 ```bash
 uv tool install git+https://github.com/xuanheng-tech/context-loader.git
 ```
 
-The repository also retains `./codex-project-context` as a direct development entry point.
+The repository also retains `./project-context` as a direct development entry point.
 
 ### Distributions
 
-From `0.1.8` the canonical PyPI distribution is **`context-loader`**. It carries the
-`context_loader` runtime and installs the `codex-project-context` console script; the CLI name,
-the JSON schema version and the public contract are unchanged by the rename.
+The sole distribution is **`context-loader`**, containing the `context_loader` runtime
+and the `project-context` console script. Version `1.0.0` breaks the CLI name and Markdown
+heading contract. The JSON schema is unchanged. Terminal and any executor call this same
+entry point with explicit repository and focus arguments; no provider adapter or private
+session state is involved.
 
-**`codex-project-context-loader`** remains published as a compatibility distribution. From
-`0.1.8` it contains no runtime and no console script and only depends on `context-loader==0.1.8`,
-so installing either name yields exactly one implementation and one CLI. Existing pins keep
-working, and releases `0.1.5`-`0.1.7` of the old name are unchanged. New integrations should
-depend on `context-loader`.
+Previously published packages and historical release records remain unchanged. The new
+release publishes no compatibility distribution or legacy executable alias.
 
 The wheel installs the runtime package only. The source distribution additionally carries
 `tool_cli_contract.json`, so a package-only consumer can pin the declared public CLI contract
@@ -91,11 +90,11 @@ automatic redaction**. Inspect the output before sharing it with a person or ext
 ## Usage
 
 ```bash
-codex-project-context --version
-codex-project-context --repo /home/user/projects/example
-codex-project-context --repo /home/user/projects/example --format markdown
-codex-project-context --repo /home/user/projects/example --format json
-codex-project-context --repo /home/user/projects/example \
+project-context --version
+project-context --repo /home/user/projects/example
+project-context --repo /home/user/projects/example --format markdown
+project-context --repo /home/user/projects/example --format json
+project-context --repo /home/user/projects/example \
   --focus "Authentication and session management" \
   --path auth/session.py
 ```
@@ -135,7 +134,7 @@ Keys are serialized in sorted order with `ensure_ascii=False`. The declared cont
   "schema_version": 1,
   "tool": {
     "name": "context-loader",
-    "version": "0.1.8"
+    "version": "1.1.0"
   },
   "repository": {
     "requested_path": "/canonical/requested/path",
@@ -164,6 +163,7 @@ Keys are serialized in sorted order with `ensure_ascii=False`. The declared cont
   ],
   "context": "the same assembled Markdown context",
   "context_sha256": "sha256-hex",
+  "statuses": [],
   "warnings": []
 }
 ```
@@ -171,15 +171,17 @@ Keys are serialized in sorted order with `ensure_ascii=False`. The declared cont
 `context_sha256` hashes the UTF-8 bytes of `context`; each `content_sha256` does the same for that
 source's `content`. `sources` contains only file bodies that actually enter the final context, in
 assembly order, after the existing newline normalization and truncation rules. `scope` distinguishes
-`repository` from `global`; version 0.1.8's fixed root-file selection currently emits only
+`repository` from `global`; version 1.0.0's fixed root-file selection currently emits only
 `repository` sources and does not add any global-file discovery.
 
+`statuses` lists machine-readable collection and render conditions that previously appeared only inside `context` Markdown: skipped or absent sources, truncated sources, unreadable directory-tree entries, sections omitted under the global output budget, and truncated working-tree or declared-command listings. Each entry has stable `code`, `subject_kind`, and `subject` fields. `sources`, `context`, and `schema_version` remain unchanged; callers can ignore `statuses` safely.
+
 The optional `selection` object is present only on a rendered `AGENTS.md` source. Its section entries
 contain heading, heading level, and fixed selection reasons; it never contains the original focus or
 target path. Existing source fields and schema version 1 remain unchanged.
 
 The JSON schema version and package version are independent: `schema_version` is currently the
-integer `1`, while `tool.version` is `0.1.8`. Callers must depend only on fields declared above.
+integer `1`, while `tool.version` is `1.1.0`. Callers must depend only on fields declared above.
 The document contains no generated time or random identifier, so unchanged input produces identical
 JSON bytes. On failure, stdout remains empty and stderr contains only a short diagnostic.
 
@@ -236,11 +238,15 @@ scan that ends before EOF still reports the characters it could not select.
 - Directory tree: 12 KiB, 300 entries, and depth 2
 - Working-tree changes: 100 paths and 4 KiB
 - Recent commits: 8
+- Git subprocess output: 16 MiB (bounded while reading)
+- Candidate file validation: 16 MiB hard read bound
 
 Truncation occurs only at complete UTF-8 and line boundaries and is marked explicitly.
-These are output/capture limits. Eligible regular files are still streamed to EOF to validate UTF-8
-and reject NUL bytes, including beyond the captured prefix; a large file can therefore take longer
-to read. Symlinked candidate files are skipped, and directory symlinks are not traversed.
+These are output/capture limits. Eligible regular files are streamed up to a hard safety limit of
+16 MiB to validate UTF-8 and reject NUL bytes, including beyond the captured prefix. A file exceeding
+this limit or failing validation fails closed and is skipped (`Skipped: unreadable.` /
+`Skipped: unsupported text encoding.`); pathological files cannot require unbounded scanning.
+Symlinked candidate files are skipped, and directory symlinks are not traversed.
 
 ## Git State Semantics
 
@@ -268,7 +274,7 @@ or candidate-file content.
 
 ## Not Included
 
-Version 0.1.8 does not provide AI summaries, project-type detection, nested `AGENTS.md` handling,
+Version 1.0.0 does not provide AI summaries, project-type detection, nested `AGENTS.md` handling,
 Memory retrieval, semantic ranking, ignore-rule parsing, plugins, profiles, caches, databases,
 network services, MCP, daemons, GUIs, CI/CD, telemetry, or automatic updates.
 

context_loader-1.1.0/context_loader/__init__.py

@@ -0,0 +1,3 @@
+"""Project Context Loader."""
+
+__version__ = "1.1.0"

{context_loader-0.1.8 → context_loader-1.1.0}/context_loader/application.py

@@ -10,9 +10,16 @@ from pathlib import Path
 
 from . import __version__
 from .collect import (
+    NOT_PRESENT,
+    SKIPPED_ENCODING,
+    SKIPPED_NOT_REGULAR,
+    SKIPPED_SYMLINK,
+    SKIPPED_UNREADABLE,
+    TRUNCATION_MARKER,
     AgentsSectionAuditEntry,
     AgentsSelectionAudit,
     AgentsSelectionInputError,
+    CollectedFile,
     ProjectContext,
     collect_project_context,
 )
@@ -22,6 +29,14 @@ from .render import render_markdown_with_details, rendered_source_contents
 JSON_SCHEMA_VERSION = 1
 TOOL_NAME = "context-loader"
 
+_STATUS_CODE_BY_MESSAGE = {
+    NOT_PRESENT: "not_present",
+    SKIPPED_SYMLINK: "skipped_symlink",
+    SKIPPED_NOT_REGULAR: "skipped_not_regular",
+    SKIPPED_ENCODING: "skipped_encoding",
+    SKIPPED_UNREADABLE: "skipped_unreadable",
+}
+
 
 @dataclass(frozen=True, slots=True)
 class ToolIdentity:
@@ -46,6 +61,15 @@ class ProjectContextSource:
     selection: AgentsSelectionAudit | None = None
 
 
+@dataclass(frozen=True, slots=True)
+class ContextStatus:
+    """One machine-readable skipped, omitted, truncated, or unreadable condition."""
+
+    code: str
+    subject_kind: str
+    subject: str
+
+
 @dataclass(frozen=True, slots=True)
 class ProjectContextResult:
     schema_version: int
@@ -55,6 +79,7 @@ class ProjectContextResult:
     context: str
     context_sha256: str
     warnings: tuple[str, ...]
+    statuses: tuple[ContextStatus, ...]
 
 
 def _text_sha256(content: str) -> str:
@@ -102,6 +127,79 @@ def _sources(
     return tuple(sources)
 
 
+def _status(code: str, subject_kind: str, subject: str) -> ContextStatus:
+    return ContextStatus(code=code, subject_kind=subject_kind, subject=subject)
+
+
+def _statuses_for_source(source: CollectedFile) -> list[ContextStatus]:
+    if source.status is not None:
+        code = _STATUS_CODE_BY_MESSAGE.get(source.status, "skipped_unreadable")
+        return [_status(code, "source", source.name)]
+    statuses: list[ContextStatus] = []
+    if source.truncated or (source.selection is not None and source.selection.truncated):
+        statuses.append(_status("truncated", "source", source.name))
+    if source.selection is not None:
+        if source.selection.parse_fallback:
+            statuses.append(_status("parse_fallback", "source", source.name))
+        if source.selection.source_scan_truncated:
+            statuses.append(_status("source_scan_truncated", "source", source.name))
+        if source.selection.index_truncated:
+            statuses.append(_status("index_truncated", "source", source.name))
+    return statuses
+
+
+def _build_statuses(
+    project: ProjectContext,
+    *,
+    omitted_sections: tuple[str, ...],
+    changes_truncated: bool,
+    commands_truncated: bool,
+    rendered_sources: tuple[ProjectContextSource, ...],
+) -> tuple[ContextStatus, ...]:
+    statuses: list[ContextStatus] = []
+    for source in (project.instructions, project.overview, *project.entry_files):
+        statuses.extend(_statuses_for_source(source))
+
+    truncated_sources = {
+        status.subject
+        for status in statuses
+        if status.subject_kind == "source" and status.code == "truncated"
+    }
+    for source in rendered_sources:
+        if TRUNCATION_MARKER not in source.content:
+            continue
+        name = source.path.name
+        if name in truncated_sources:
+            continue
+        statuses.append(_status("truncated", "source", name))
+        truncated_sources.add(name)
+
+    if commands_truncated:
+        statuses.append(_status("truncated", "commands", "Declared Commands"))
+    if changes_truncated:
+        statuses.append(_status("truncated", "changes", "Working Tree Changes"))
+
+    if project.directory_tree.truncated:
+        statuses.append(_status("truncated", "tree", "Directory Tree"))
+    for entry in project.directory_tree.entries:
+        if entry.kind == "unreadable_directory":
+            subject = entry.path if entry.path else "."
+            statuses.append(_status("unreadable", "tree_entry", subject))
+
+    for title in omitted_sections:
+        statuses.append(_status("section_omitted", "section", title))
+
+    deduped: list[ContextStatus] = []
+    seen: set[tuple[str, str, str]] = set()
+    for status in statuses:
+        key = (status.code, status.subject_kind, status.subject)
+        if key in seen:
+            continue
+        seen.add(key)
+        deduped.append(status)
+    return tuple(deduped)
+
+
 def load_project_context(
     repo: str | os.PathLike[str],
     *,
@@ -120,6 +218,14 @@ def load_project_context(
         raise ContextLoaderError(str(exc), exit_code=2) from None
     rendered = render_markdown_with_details(state, project)
     context = rendered.output.decode("utf-8")
+    sources = _sources(state.repository, project, rendered.included_sections)
+    statuses = _build_statuses(
+        project,
+        omitted_sections=rendered.omitted_sections,
+        changes_truncated=rendered.changes_truncated,
+        commands_truncated=rendered.commands_truncated,
+        rendered_sources=sources,
+    )
     return ProjectContextResult(
         schema_version=JSON_SCHEMA_VERSION,
         tool=ToolIdentity(name=TOOL_NAME, version=__version__),
@@ -127,10 +233,11 @@ def load_project_context(
             requested_path=location.requested_path,
             canonical_root=location.canonical_root,
         ),
-        sources=_sources(state.repository, project, rendered.included_sections),
+        sources=sources,
         context=context,
         context_sha256=hashlib.sha256(rendered.output).hexdigest(),
         warnings=(),
+        statuses=statuses,
     )
 
 
@@ -157,6 +264,14 @@ def _selection_document(selection: AgentsSelectionAudit) -> dict[str, object]:
     }
 
 
+def _status_document(status: ContextStatus) -> dict[str, object]:
+    return {
+        "code": status.code,
+        "subject": status.subject,
+        "subject_kind": status.subject_kind,
+    }
+
+
 def _source_document(source: ProjectContextSource) -> dict[str, object]:
     document: dict[str, object] = {
         "ordinal": source.ordinal,
@@ -184,6 +299,7 @@ def render_json(result: ProjectContextResult) -> bytes:
             "canonical_root": os.fspath(result.repository.canonical_root),
         },
         "sources": [_source_document(source) for source in result.sources],
+        "statuses": [_status_document(status) for status in result.statuses],
         "context": result.context,
         "context_sha256": result.context_sha256,
         "warnings": list(result.warnings),

{context_loader-0.1.8 → context_loader-1.1.0}/context_loader/cli.py

@@ -1,4 +1,4 @@
-"""Command-line interface for deterministic Codex project context."""
+"""Command-line interface for deterministic project context."""
 
 from __future__ import annotations
 
@@ -19,7 +19,7 @@ class SafeArgumentParser(argparse.ArgumentParser):
 
 def _parser() -> argparse.ArgumentParser:
     parser = SafeArgumentParser(
-        prog="codex-project-context",
+        prog="project-context",
         description="Render deterministic local Git context as Markdown or JSON.",
         allow_abbrev=False,
     )

{context_loader-0.1.8 → context_loader-1.1.0}/context_loader/collect.py

@@ -26,6 +26,8 @@ DECLARED_COMMANDS_LIMIT_BYTES = 8 * 1024
 DIRECTORY_TREE_LIMIT_BYTES = 12 * 1024
 DIRECTORY_TREE_MAX_ITEMS = 300
 DIRECTORY_TREE_MAX_DEPTH = 2
+FILE_SCAN_LIMIT_BYTES = 16 * 1024 * 1024
+MAX_FILE_SCAN_BYTES = FILE_SCAN_LIMIT_BYTES
 TRUNCATION_MARKER = "… truncated by context-loader …"
 
 NOT_PRESENT = "Not present."
@@ -705,11 +707,15 @@ def _read_validated_text(file_descriptor: int, limit: int) -> tuple[str, bool, s
             source_characters += 1
             append(character)
 
+    total_bytes_read = 0
     try:
         while True:
             raw = os.read(file_descriptor, 64 * 1024)
             if not raw:
                 break
+            total_bytes_read += len(raw)
+            if total_bytes_read > FILE_SCAN_LIMIT_BYTES:
+                return "", False, SKIPPED_UNREADABLE, 0
             if b"\0" in raw:
                 return "", False, SKIPPED_ENCODING, 0
             consume(decoder.decode(raw, final=False))
@@ -743,6 +749,8 @@ def _collect_root_file(root: Path, name: str, language: str, limit: int) -> Coll
         return CollectedFile(name, language, SKIPPED_SYMLINK)
     if not stat.S_ISREG(metadata.st_mode):
         return CollectedFile(name, language, SKIPPED_NOT_REGULAR)
+    if metadata.st_size > FILE_SCAN_LIMIT_BYTES:
+        return CollectedFile(name, language, SKIPPED_UNREADABLE)
 
     flags = os.O_RDONLY | os.O_CLOEXEC | os.O_NONBLOCK | getattr(os, "O_NOFOLLOW", 0)
     try:
@@ -752,8 +760,11 @@ def _collect_root_file(root: Path, name: str, language: str, limit: int) -> Coll
             return CollectedFile(name, language, SKIPPED_SYMLINK)
         return CollectedFile(name, language, SKIPPED_UNREADABLE)
     try:
-        if not stat.S_ISREG(os.fstat(file_descriptor).st_mode):
+        stat_result = os.fstat(file_descriptor)
+        if not stat.S_ISREG(stat_result.st_mode):
             return CollectedFile(name, language, SKIPPED_NOT_REGULAR)
+        if stat_result.st_size > FILE_SCAN_LIMIT_BYTES:
+            return CollectedFile(name, language, SKIPPED_UNREADABLE)
         content, truncated, error_status, source_characters = _read_validated_text(
             file_descriptor, limit
         )

{context_loader-0.1.8 → context_loader-1.1.0}/context_loader/git.py

@@ -4,7 +4,9 @@ from __future__ import annotations
 
 import os
 import re
+import selectors
 import subprocess
+import time
 from dataclasses import dataclass
 from pathlib import Path
 from types import MappingProxyType
@@ -108,25 +110,81 @@ class RepositoryLocation:
     canonical_root: Path
 
 
+def _read_bounded_process_stdout(
+    proc: subprocess.Popen[bytes],
+    max_bytes: int,
+    timeout_seconds: float,
+) -> bytes:
+    deadline = time.monotonic() + timeout_seconds
+    buffer = bytearray()
+    assert proc.stdout is not None
+    os.set_blocking(proc.stdout.fileno(), False)
+    selector = selectors.DefaultSelector()
+    selector.register(proc.stdout, selectors.EVENT_READ)
+    try:
+        while True:
+            remaining = deadline - time.monotonic()
+            if remaining <= 0:
+                proc.kill()
+                proc.wait()
+                raise subprocess.TimeoutExpired(proc.args, timeout_seconds)
+            events = selector.select(timeout=max(0.0, remaining))
+            if not events:
+                proc.kill()
+                proc.wait()
+                raise subprocess.TimeoutExpired(proc.args, timeout_seconds)
+            try:
+                chunk = os.read(proc.stdout.fileno(), 64 * 1024)
+            except BlockingIOError:
+                continue
+            if not chunk:
+                break
+            buffer.extend(chunk)
+            if len(buffer) > max_bytes:
+                proc.kill()
+                proc.wait()
+                raise ContextLoaderError("repository Git output exceeded the safety limit")
+    except Exception:
+        try:
+            proc.kill()
+            proc.wait()
+        except OSError:
+            pass
+        raise
+    finally:
+        selector.close()
+
+    remaining = max(0.0, deadline - time.monotonic())
+    try:
+        proc.wait(timeout=remaining)
+    except subprocess.TimeoutExpired:
+        proc.kill()
+        proc.wait()
+        raise
+    return bytes(buffer)
+
+
 def _run_git(repo: Path, arguments: tuple[str, ...], *, check: bool = True) -> GitResult:
     argv = [GIT_EXECUTABLE, *_GIT_PREFIX, "-C", os.fspath(repo), *arguments]
     try:
-        completed = subprocess.run(  # noqa: S603 - executable and every command are fixed here.
+        with subprocess.Popen(  # noqa: S603 - executable and every command are fixed here.
             argv,
             cwd="/",
             env=dict(_GIT_ENVIRONMENT),
             stdin=subprocess.DEVNULL,
-            capture_output=True,
-            timeout=GIT_TIMEOUT_SECONDS,
-            check=False,
-        )
+            stdout=subprocess.PIPE,
+            stderr=subprocess.DEVNULL,
+        ) as proc:
+            stdout = _read_bounded_process_stdout(proc, MAX_GIT_OUTPUT_BYTES, GIT_TIMEOUT_SECONDS)
+            returncode = proc.returncode
+    except ContextLoaderError:
+        raise
     except (OSError, subprocess.TimeoutExpired) as exc:
         raise ContextLoaderError("unable to read repository Git state") from exc
-    if len(completed.stdout) > MAX_GIT_OUTPUT_BYTES:
-        raise ContextLoaderError("repository Git output exceeded the safety limit")
-    if check and completed.returncode != 0:
+
+    if check and returncode != 0:
         raise ContextLoaderError("unable to read repository Git state")
-    return GitResult(completed.stdout, completed.returncode)
+    return GitResult(stdout, returncode)
 
 
 def _without_one_line_ending(raw: bytes) -> bytes:

{context_loader-0.1.8 → context_loader-1.1.0}/context_loader/render.py

@@ -34,6 +34,9 @@ GLOBAL_OMISSION = "Omitted: global output limit reached."
 class MarkdownRender:
     output: bytes
     included_sections: tuple[str, ...]
+    omitted_sections: tuple[str, ...]
+    changes_truncated: bool
+    commands_truncated: bool
 
 
 def _display(value: str) -> str:
@@ -173,10 +176,12 @@ def _command_line(command: DeclaredCommand) -> str:
     return f"- `{invocation}` → `{_display(command.target)}`"
 
 
-def _bounded_lines(lines: list[str], limit: int, *, already_truncated: bool = False) -> str:
+def _bounded_lines(
+    lines: list[str], limit: int, *, already_truncated: bool = False
+) -> tuple[str, bool]:
     complete = "\n".join(lines)
     if not already_truncated and len(complete.encode()) <= limit:
-        return complete
+        return complete, False
     marker_size = len(TRUNCATION_MARKER.encode())
     budget = max(0, limit - marker_size - 1)
     selected: list[str] = []
@@ -188,20 +193,15 @@ def _bounded_lines(lines: list[str], limit: int, *, already_truncated: bool = Fa
         selected.append(line)
         used += size
     selected.append(TRUNCATION_MARKER)
-    return "\n".join(selected)
+    return "\n".join(selected), True
 
 
-def _render_commands(commands: tuple[DeclaredCommand, ...]) -> str:
+def _render_commands(commands: tuple[DeclaredCommand, ...]) -> tuple[str, bool]:
     lines = [_command_line(command) for command in commands]
     if not lines:
         lines.append("No supported command declarations found.")
-    return "\n".join(
-        (
-            "## Declared Commands",
-            "",
-            _bounded_lines(lines, DECLARED_COMMANDS_LIMIT_BYTES),
-        )
-    )
+    body, truncated = _bounded_lines(lines, DECLARED_COMMANDS_LIMIT_BYTES)
+    return "\n".join(("## Declared Commands", "", body)), truncated
 
 
 def _entry_body(source: CollectedFile, remaining: int) -> tuple[str | None, int]:
@@ -276,7 +276,7 @@ def _render_directory_tree(tree: DirectoryTree) -> str:
     lines = [_tree_line(entry) for entry in tree.entries]
     if not lines:
         lines.append("No entries.")
-    body = _bounded_lines(
+    body, _truncated = _bounded_lines(
         lines,
         DIRECTORY_TREE_LIMIT_BYTES,
         already_truncated=tree.truncated,
@@ -292,12 +292,14 @@ def render_markdown_with_details(state: RepositoryState, project: ProjectContext
     """Render all sections in fixed order without exceeding the global byte limit."""
     header = "\n".join(
         (
-            "# Codex Project Context",
+            "# Project Context",
             "",
             f"- Schema: `{SCHEMA}`",
             f"- Repository: `{_display(os.fspath(state.repository))}`",
         )
     )
+    commands_section, commands_truncated = _render_commands(project.commands)
+    changes_truncated = _change_lines(state)[1]
     sections: list[tuple[str, str | None]] = [
         ("Git State", _render_git_state(state)),
         (
@@ -312,7 +314,7 @@ def render_markdown_with_details(state: RepositoryState, project: ProjectContext
             "Project Overview",
             _render_source_section("Project Overview", project.overview, README_LIMIT_BYTES),
         ),
-        ("Declared Commands", _render_commands(project.commands)),
+        ("Declared Commands", commands_section),
         ("Project Entry Files", _render_entry_files(project.entry_files)),
         ("Recent Commits", _render_recent_commits(state.commits)),
         ("Directory Tree", _render_directory_tree(project.directory_tree)),
@@ -320,6 +322,7 @@ def render_markdown_with_details(state: RepositoryState, project: ProjectContext
 
     rendered_parts = [header.encode()]
     included_sections: list[str] = []
+    omitted_sections: list[str] = []
     omission_started = False
     for index, (title, full_section) in enumerate(sections):
         omission = _omitted_section(title).encode()
@@ -330,6 +333,7 @@ def render_markdown_with_details(state: RepositoryState, project: ProjectContext
         if omission_started or full_section is None:
             selected = omission
             omission_started = True
+            omitted_sections.append(title)
         else:
             candidate = full_section.encode()
             tentative = b"\n\n".join((*rendered_parts, candidate, *future_omissions)) + b"\n"
@@ -339,12 +343,19 @@ def render_markdown_with_details(state: RepositoryState, project: ProjectContext
             else:
                 selected = omission
                 omission_started = True
+                omitted_sections.append(title)
         rendered_parts.append(selected)
 
     output = b"\n\n".join(rendered_parts) + b"\n"
     if len(output) > GLOBAL_OUTPUT_LIMIT_BYTES:
         raise RuntimeError("global context output limit could not be satisfied")
-    return MarkdownRender(output=output, included_sections=tuple(included_sections))
+    return MarkdownRender(
+        output=output,
+        included_sections=tuple(included_sections),
+        omitted_sections=tuple(omitted_sections),
+        changes_truncated=changes_truncated,
+        commands_truncated=commands_truncated,
+    )
 
 
 def rendered_source_contents(

{context_loader-0.1.8 → context_loader-1.1.0}/pyproject.toml

@@ -4,8 +4,8 @@ build-backend = "uv_build"
 
 [project]
 name = "context-loader"
-version = "0.1.8"
-description = "Render deterministic local Git context for Codex"
+version = "1.1.0"
+description = "Render deterministic local Git context for any executor"
 readme = "README.md"
 requires-python = ">=3.12,<3.13"
 dependencies = []
@@ -14,7 +14,7 @@ dependencies = []
 file = "LICENSE"
 
 [project.scripts]
-codex-project-context = "context_loader.cli:main"
+project-context = "context_loader.cli:main"
 
 [project.urls]
 Repository = "https://github.com/xuanheng-tech/context-loader"

{context_loader-0.1.8 → context_loader-1.1.0}/pyproject.toml.orig

@@ -4,15 +4,15 @@ build-backend = "uv_build"
 
 [project]
 name = "context-loader"
-version = "0.1.8"
-description = "Render deterministic local Git context for Codex"
+version = "1.1.0"
+description = "Render deterministic local Git context for any executor"
 license = { file = "LICENSE" }
 readme = "README.md"
 requires-python = ">=3.12,<3.13"
 dependencies = []
 
 [project.scripts]
-codex-project-context = "context_loader.cli:main"
+project-context = "context_loader.cli:main"
 
 [project.urls]
 Repository = "https://github.com/xuanheng-tech/context-loader"

{context_loader-0.1.8 → context_loader-1.1.0}/tool_cli_contract.json

@@ -9,7 +9,7 @@
         "--version",
         "--help"
       ],
-      "name": "codex-project-context",
+      "name": "project-context",
       "operation_class": "read_only",
       "preconditions": [
         "repo_is_absolute_git_worktree",
@@ -24,7 +24,7 @@
       ]
     }
   ],
-  "contract_version": 1,
+  "contract_version": 2,
   "flags": {
     "format": [
       "markdown",
@@ -44,5 +44,5 @@
   ],
   "schema_version": 1,
   "tool_name": "context-loader",
-  "tool_version": "0.1.8"
+  "tool_version": "1.1.0"
 }

context_loader-0.1.8/context_loader/__init__.py

@@ -1,3 +0,0 @@
-"""Codex Project Context Loader."""
-
-__version__ = "0.1.8"