codex2claude 0.1.2__tar.gz

codex2claude-0.1.2/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

codex2claude-0.1.2/PKG-INFO

@@ -0,0 +1,300 @@
+Metadata-Version: 2.4
+Name: codex2claude
+Version: 0.1.2
+Summary: Local Codex-to-Claude bridge CLI
+Author: OpenAI Codex
+License-Expression: MIT
+Project-URL: Homepage, https://github.com/Housetan218/codex2claude
+Project-URL: Repository, https://github.com/Housetan218/codex2claude
+Project-URL: Issues, https://github.com/Housetan218/codex2claude/issues
+Project-URL: Releases, https://github.com/Housetan218/codex2claude/releases
+Keywords: codex,claude,cli,bridge,agent
+Classifier: Development Status :: 4 - Beta
+Classifier: Environment :: Console
+Classifier: Intended Audience :: Developers
+Classifier: Operating System :: MacOS
+Classifier: Operating System :: POSIX
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Topic :: Software Development :: Build Tools
+Classifier: Topic :: Utilities
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Dynamic: license-file
+
+# codex2claude
+
+`codex2claude` is a local one-way bridge from Codex to Claude.
+
+It provides:
+
+- a reusable Python CLI bridge
+- Claude session persistence via native `session_id`
+- deterministic per-thread locking
+- a thin Codex skill wrapper surface
+- no non-stdlib Python runtime dependency inside the bridge
+
+Current implementation target:
+
+- macOS / POSIX environments with Python 3 and local Claude CLI access
+- not designed for Windows in its current `fcntl`-based form
+
+## Current Status
+
+Implemented and locally verified on this machine:
+
+- `ask`
+- `status`
+- `forget`
+- `gc`
+- automatic resume via stored Claude `session_id`
+- same-thread lock conflict handling
+
+Fresh verification completed during implementation:
+
+- `PYTHONPATH=bridge python3 -m unittest discover -s tests -p 'test_*.py' -v`
+- real Claude new-session smoke
+- real Claude resume smoke
+
+## Install
+
+```bash
+python3 -m venv .venv
+source .venv/bin/activate
+python3 -m pip install -e .
+```
+
+When a PyPI release is available, install with:
+
+```bash
+python3 -m pip install codex2claude
+```
+
+If you do not want to install it yet, you can run it directly:
+
+```bash
+PYTHONPATH=bridge python3 -m codex2claude ask --prompt "Reply with ok only"
+```
+
+After editable install, both of these work:
+
+```bash
+python -m codex2claude --help
+codex2claude --help
+```
+
+## Usage
+
+Ask Claude in the current workspace thread:
+
+```bash
+codex2claude ask --prompt "Review this design" --workspace "$PWD"
+```
+
+Use a named thread:
+
+```bash
+codex2claude ask --prompt "Continue the design review" --workspace "$PWD" --thread design
+```
+
+Force a fresh Claude session:
+
+```bash
+codex2claude ask --prompt "Start over" --workspace "$PWD" --new
+```
+
+Inspect stored state:
+
+```bash
+codex2claude status --workspace "$PWD"
+```
+
+Forget the current thread:
+
+```bash
+codex2claude forget --workspace "$PWD"
+```
+
+Remove stale thread files:
+
+```bash
+codex2claude gc --max-age-days 7
+```
+
+## Codex Skill
+
+The Codex-facing wrapper lives at:
+
+```text
+skills/codex-to-claude/SKILL.md
+```
+
+The skill should stay thin. It should only:
+
+- collect the user prompt
+- choose default thread, named thread, or `--new`
+- invoke `codex2claude`
+- return stdout or surface stderr
+
+It should not own Claude JSON parsing, session files, or retries.
+
+Common trigger phrases that should explicitly steer Codex toward this skill:
+
+- `Use the codex-to-claude skill`
+- `ask Claude about this`
+- `send this to Claude`
+- `let Claude review this`
+- `ask cc`
+- `ask cc about this`
+- `cc review this`
+- `cc 怎么看`
+- `cc 觉得呢`
+- `cc 能帮忙看下吗`
+- `问问cc`
+- `问下Claude`
+- `Claude 怎么看`
+- `Claude 能看下吗`
+- `让cc帮忙看看`
+- `cc review下`
+- `cc check一下`
+- `give this to cc`
+- `给 Claude 看看`
+- `让 Claude 看一下`
+- `发给 Claude`
+- `给 cc 看看`
+- `让 cc review 一下`
+- `发给 cc`
+
+For stability, prefer phrases where `cc` appears with an action like ask, review, check, or look. Avoid relying on bare `cc` by itself.
+
+## Triggering
+
+Recommended everyday trigger phrases:
+
+- `给cc看看这个`
+- `问问cc`
+- `cc怎么看`
+- `让cc review一下`
+- `给Claude看看`
+
+More explicit variants:
+
+- `Use the codex-to-claude skill`
+- `ask Claude about this`
+- `ask cc`
+- `cc review this`
+- `给 cc 看看`
+- `问下Claude`
+
+## Activation Scope
+
+These paths are confirmed to work:
+
+- new Codex sessions started after the skill was installed
+- `codex exec` runs started after the skill was installed
+- direct `codex2claude` CLI usage
+
+Do not assume already-open Codex sessions will hot-reload newly installed or updated skills.
+
+If you changed trigger phrases or installed the skill during an existing session, restart Codex or open a fresh session before testing.
+
+## Troubleshooting
+
+If a trigger phrase does not route to Claude:
+
+1. Start a new Codex session.
+2. Test with a high-signal phrase such as `问问cc：请只回复 ok`.
+3. If needed, use the most explicit form: `Use the codex-to-claude skill. Ask Claude: ...`
+4. Verify the bridge directly with `codex2claude ask --prompt "Reply with ok only" --workspace "$PWD"`.
+
+If direct CLI usage works but a natural-language trigger does not, the issue is skill discovery in that session, not the bridge itself.
+
+## Threading Model
+
+By default, one workspace maps to one Claude thread.
+
+Use `--thread <name>` to split conversations inside the same repo:
+
+```bash
+codex2claude ask --prompt "Review API design" --workspace "$PWD" --thread api
+codex2claude ask --prompt "Review docs tone" --workspace "$PWD" --thread docs
+```
+
+Use `--new` when you want a fresh Claude conversation for the selected thread key.
+
+## Environment
+
+- `CODEX2CLAUDE_CLAUDE_BIN`: override the Claude executable path
+- `CODEX2CLAUDE_HOME`: override the bridge state root without changing your real shell `HOME`
+- `CODEX2CLAUDE_RUN_REAL=1`: enable opt-in real Claude integration tests
+
+## Test
+
+Run the main test suite:
+
+```bash
+PYTHONPATH=bridge python3 -m unittest discover -s tests -p 'test_*.py' -v
+```
+
+Run the opt-in real Claude smoke test:
+
+```bash
+PYTHONPATH=bridge CODEX2CLAUDE_RUN_REAL=1 python3 -m unittest tests.test_real_claude_cli -v
+```
+
+The real-Claude test is opt-in because it spends actual Claude usage and requires local auth.
+
+## Exit Codes
+
+- `0`: success
+- `1`: Claude invocation failure or generic bridge failure
+- `2`: Claude timeout
+- `3`: same-thread lock conflict
+- `4`: invalid arguments
+- `5`: corrupted state or persistence failure
+
+## State Layout
+
+```text
+~/.codex/codex2claude/
+  threads/
+  runs/
+  logs/
+```
+
+Important files:
+
+- `threads/<thread_key>.json`: current thread state and stored Claude `session_id`
+- `runs/<thread_key>/...json`: per-run artifacts
+- `logs/bridge.log`: append-only bridge events
+
+## Current Scope
+
+This version is intentionally one-way only:
+
+- Codex initiates
+- Claude replies
+- bridge stores Claude `session_id`
+- follow-up turns use native `claude --resume`
+
+Bidirectional agent protocols are out of scope for v1.
+
+## References
+
+- Design: `docs/superpowers/specs/2026-03-28-codex-to-claude-design.md`
+- Plan: `docs/superpowers/plans/2026-03-28-codex-to-claude-v1.md`
+
+## Contributing
+
+See [CONTRIBUTING.md](./CONTRIBUTING.md) for development, verification, and release steps.
+
+## CI And Releases
+
+GitHub Actions runs unit tests and packaging checks on pushes and pull requests.
+
+PyPI publishing is wired through `.github/workflows/release.yml` and is intended to use PyPI Trusted Publishing from GitHub Actions. Repository maintainers still need to configure the matching trusted publisher entry on PyPI before tag-based publishing can succeed.

codex2claude-0.1.2/README.md

@@ -0,0 +1,271 @@
+# codex2claude
+
+`codex2claude` is a local one-way bridge from Codex to Claude.
+
+It provides:
+
+- a reusable Python CLI bridge
+- Claude session persistence via native `session_id`
+- deterministic per-thread locking
+- a thin Codex skill wrapper surface
+- no non-stdlib Python runtime dependency inside the bridge
+
+Current implementation target:
+
+- macOS / POSIX environments with Python 3 and local Claude CLI access
+- not designed for Windows in its current `fcntl`-based form
+
+## Current Status
+
+Implemented and locally verified on this machine:
+
+- `ask`
+- `status`
+- `forget`
+- `gc`
+- automatic resume via stored Claude `session_id`
+- same-thread lock conflict handling
+
+Fresh verification completed during implementation:
+
+- `PYTHONPATH=bridge python3 -m unittest discover -s tests -p 'test_*.py' -v`
+- real Claude new-session smoke
+- real Claude resume smoke
+
+## Install
+
+```bash
+python3 -m venv .venv
+source .venv/bin/activate
+python3 -m pip install -e .
+```
+
+When a PyPI release is available, install with:
+
+```bash
+python3 -m pip install codex2claude
+```
+
+If you do not want to install it yet, you can run it directly:
+
+```bash
+PYTHONPATH=bridge python3 -m codex2claude ask --prompt "Reply with ok only"
+```
+
+After editable install, both of these work:
+
+```bash
+python -m codex2claude --help
+codex2claude --help
+```
+
+## Usage
+
+Ask Claude in the current workspace thread:
+
+```bash
+codex2claude ask --prompt "Review this design" --workspace "$PWD"
+```
+
+Use a named thread:
+
+```bash
+codex2claude ask --prompt "Continue the design review" --workspace "$PWD" --thread design
+```
+
+Force a fresh Claude session:
+
+```bash
+codex2claude ask --prompt "Start over" --workspace "$PWD" --new
+```
+
+Inspect stored state:
+
+```bash
+codex2claude status --workspace "$PWD"
+```
+
+Forget the current thread:
+
+```bash
+codex2claude forget --workspace "$PWD"
+```
+
+Remove stale thread files:
+
+```bash
+codex2claude gc --max-age-days 7
+```
+
+## Codex Skill
+
+The Codex-facing wrapper lives at:
+
+```text
+skills/codex-to-claude/SKILL.md
+```
+
+The skill should stay thin. It should only:
+
+- collect the user prompt
+- choose default thread, named thread, or `--new`
+- invoke `codex2claude`
+- return stdout or surface stderr
+
+It should not own Claude JSON parsing, session files, or retries.
+
+Common trigger phrases that should explicitly steer Codex toward this skill:
+
+- `Use the codex-to-claude skill`
+- `ask Claude about this`
+- `send this to Claude`
+- `let Claude review this`
+- `ask cc`
+- `ask cc about this`
+- `cc review this`
+- `cc 怎么看`
+- `cc 觉得呢`
+- `cc 能帮忙看下吗`
+- `问问cc`
+- `问下Claude`
+- `Claude 怎么看`
+- `Claude 能看下吗`
+- `让cc帮忙看看`
+- `cc review下`
+- `cc check一下`
+- `give this to cc`
+- `给 Claude 看看`
+- `让 Claude 看一下`
+- `发给 Claude`
+- `给 cc 看看`
+- `让 cc review 一下`
+- `发给 cc`
+
+For stability, prefer phrases where `cc` appears with an action like ask, review, check, or look. Avoid relying on bare `cc` by itself.
+
+## Triggering
+
+Recommended everyday trigger phrases:
+
+- `给cc看看这个`
+- `问问cc`
+- `cc怎么看`
+- `让cc review一下`
+- `给Claude看看`
+
+More explicit variants:
+
+- `Use the codex-to-claude skill`
+- `ask Claude about this`
+- `ask cc`
+- `cc review this`
+- `给 cc 看看`
+- `问下Claude`
+
+## Activation Scope
+
+These paths are confirmed to work:
+
+- new Codex sessions started after the skill was installed
+- `codex exec` runs started after the skill was installed
+- direct `codex2claude` CLI usage
+
+Do not assume already-open Codex sessions will hot-reload newly installed or updated skills.
+
+If you changed trigger phrases or installed the skill during an existing session, restart Codex or open a fresh session before testing.
+
+## Troubleshooting
+
+If a trigger phrase does not route to Claude:
+
+1. Start a new Codex session.
+2. Test with a high-signal phrase such as `问问cc：请只回复 ok`.
+3. If needed, use the most explicit form: `Use the codex-to-claude skill. Ask Claude: ...`
+4. Verify the bridge directly with `codex2claude ask --prompt "Reply with ok only" --workspace "$PWD"`.
+
+If direct CLI usage works but a natural-language trigger does not, the issue is skill discovery in that session, not the bridge itself.
+
+## Threading Model
+
+By default, one workspace maps to one Claude thread.
+
+Use `--thread <name>` to split conversations inside the same repo:
+
+```bash
+codex2claude ask --prompt "Review API design" --workspace "$PWD" --thread api
+codex2claude ask --prompt "Review docs tone" --workspace "$PWD" --thread docs
+```
+
+Use `--new` when you want a fresh Claude conversation for the selected thread key.
+
+## Environment
+
+- `CODEX2CLAUDE_CLAUDE_BIN`: override the Claude executable path
+- `CODEX2CLAUDE_HOME`: override the bridge state root without changing your real shell `HOME`
+- `CODEX2CLAUDE_RUN_REAL=1`: enable opt-in real Claude integration tests
+
+## Test
+
+Run the main test suite:
+
+```bash
+PYTHONPATH=bridge python3 -m unittest discover -s tests -p 'test_*.py' -v
+```
+
+Run the opt-in real Claude smoke test:
+
+```bash
+PYTHONPATH=bridge CODEX2CLAUDE_RUN_REAL=1 python3 -m unittest tests.test_real_claude_cli -v
+```
+
+The real-Claude test is opt-in because it spends actual Claude usage and requires local auth.
+
+## Exit Codes
+
+- `0`: success
+- `1`: Claude invocation failure or generic bridge failure
+- `2`: Claude timeout
+- `3`: same-thread lock conflict
+- `4`: invalid arguments
+- `5`: corrupted state or persistence failure
+
+## State Layout
+
+```text
+~/.codex/codex2claude/
+  threads/
+  runs/
+  logs/
+```
+
+Important files:
+
+- `threads/<thread_key>.json`: current thread state and stored Claude `session_id`
+- `runs/<thread_key>/...json`: per-run artifacts
+- `logs/bridge.log`: append-only bridge events
+
+## Current Scope
+
+This version is intentionally one-way only:
+
+- Codex initiates
+- Claude replies
+- bridge stores Claude `session_id`
+- follow-up turns use native `claude --resume`
+
+Bidirectional agent protocols are out of scope for v1.
+
+## References
+
+- Design: `docs/superpowers/specs/2026-03-28-codex-to-claude-design.md`
+- Plan: `docs/superpowers/plans/2026-03-28-codex-to-claude-v1.md`
+
+## Contributing
+
+See [CONTRIBUTING.md](./CONTRIBUTING.md) for development, verification, and release steps.
+
+## CI And Releases
+
+GitHub Actions runs unit tests and packaging checks on pushes and pull requests.
+
+PyPI publishing is wired through `.github/workflows/release.yml` and is intended to use PyPI Trusted Publishing from GitHub Actions. Repository maintainers still need to configure the matching trusted publisher entry on PyPI before tag-based publishing can succeed.

codex2claude-0.1.2/bridge/codex2claude/__init__.py

@@ -0,0 +1,3 @@
+from .version import __version__
+
+__all__ = ["__version__"]

codex2claude-0.1.2/bridge/codex2claude/__main__.py

@@ -0,0 +1,5 @@
+from .cli import main
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

codex2claude-0.1.2/bridge/codex2claude/claude_cli.py

@@ -0,0 +1,96 @@
+from __future__ import annotations
+
+import json
+import subprocess
+from dataclasses import dataclass
+
+from .errors import ClaudeInvocationError, ClaudeTimeoutError, StateCorruptionError
+
+
+@dataclass(slots=True)
+class ClaudeResult:
+    session_id: str | None
+    result_text: str
+    raw_payload: dict[str, object]
+    stderr_text: str
+
+
+def build_claude_command(prompt: str, session_id: str | None, claude_bin: str = "claude") -> list[str]:
+    command = [claude_bin]
+    if session_id:
+        command.extend(["--resume", session_id])
+    command.extend(["-p", prompt, "--output-format", "json"])
+    return command
+
+
+def parse_claude_response(stdout: str, stderr: str = "") -> ClaudeResult:
+    try:
+        payload = json.loads(stdout)
+    except json.JSONDecodeError as exc:
+        raise StateCorruptionError("Claude returned malformed JSON") from exc
+
+    result_text = payload.get("result")
+    if not isinstance(result_text, str):
+        raise StateCorruptionError("Claude JSON missing string result")
+
+    if payload.get("is_error") is True:
+        raise ClaudeInvocationError(result_text)
+
+    session_id = payload.get("session_id")
+    if session_id is not None and not isinstance(session_id, str):
+        raise StateCorruptionError("Claude JSON returned invalid session_id")
+    if isinstance(session_id, str) and not session_id:
+        raise StateCorruptionError("Claude JSON returned empty session_id")
+
+    return ClaudeResult(
+        session_id=session_id,
+        result_text=result_text,
+        raw_payload=payload,
+        stderr_text=stderr,
+    )
+
+
+def invoke_claude(
+    prompt: str,
+    session_id: str | None,
+    timeout_seconds: int,
+    cwd: str | None = None,
+    claude_bin: str = "claude",
+) -> ClaudeResult:
+    command = build_claude_command(prompt=prompt, session_id=session_id, claude_bin=claude_bin)
+    try:
+        completed = subprocess.run(
+            command,
+            check=False,
+            capture_output=True,
+            text=True,
+            timeout=timeout_seconds,
+            cwd=cwd,
+        )
+    except FileNotFoundError as exc:
+        raise ClaudeInvocationError(f"Claude CLI not found: {claude_bin}") from exc
+    except subprocess.TimeoutExpired as exc:
+        raise ClaudeTimeoutError(f"Claude timed out after {timeout_seconds}s") from exc
+
+    if completed.returncode != 0:
+        stderr = completed.stderr.strip() or completed.stdout.strip() or "unknown Claude failure"
+        raise ClaudeInvocationError(stderr)
+
+    return parse_claude_response(completed.stdout, completed.stderr)
+
+
+def read_claude_version(claude_bin: str = "claude") -> str | None:
+    try:
+        completed = subprocess.run(
+            [claude_bin, "--version"],
+            check=False,
+            capture_output=True,
+            text=True,
+            timeout=20,
+        )
+    except (FileNotFoundError, subprocess.TimeoutExpired):
+        return None
+    if completed.returncode != 0:
+        return None
+    version = completed.stdout.strip() or completed.stderr.strip()
+    return version or None