stay-safe 0.2.3__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- stay_safe-0.2.3/LICENSE +75 -0
- stay_safe-0.2.3/MANIFEST.in +1 -0
- stay_safe-0.2.3/PKG-INFO +103 -0
- stay_safe-0.2.3/README.md +91 -0
- stay_safe-0.2.3/harnesses/codex.md +175 -0
- stay_safe-0.2.3/incidents.md +159 -0
- stay_safe-0.2.3/pyproject.toml +22 -0
- stay_safe-0.2.3/setup.cfg +4 -0
- stay_safe-0.2.3/setup.py +47 -0
- stay_safe-0.2.3/stay_safe/__init__.py +35 -0
- stay_safe-0.2.3/stay_safe/__main__.py +20 -0
- stay_safe-0.2.3/stay_safe/cli.py +156 -0
- stay_safe-0.2.3/stay_safe/codex.py +413 -0
- stay_safe-0.2.3/stay_safe/codex_rules.py +80 -0
- stay_safe-0.2.3/stay_safe/provenance.py +127 -0
- stay_safe-0.2.3/stay_safe/release.txt +1 -0
- stay_safe-0.2.3/stay_safe/report.py +1066 -0
- stay_safe-0.2.3/stay_safe/rules.py +53 -0
- stay_safe-0.2.3/stay_safe/save.py +33 -0
- stay_safe-0.2.3/stay_safe.egg-info/PKG-INFO +103 -0
- stay_safe-0.2.3/stay_safe.egg-info/SOURCES.txt +22 -0
- stay_safe-0.2.3/stay_safe.egg-info/dependency_links.txt +1 -0
- stay_safe-0.2.3/stay_safe.egg-info/entry_points.txt +2 -0
- stay_safe-0.2.3/stay_safe.egg-info/top_level.txt +1 -0
stay_safe-0.2.3/LICENSE
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
1
|
+
Copyright (c) 2026 chalkbadger
|
|
2
|
+
|
|
3
|
+
Apache License
|
|
4
|
+
Version 2.0, January 2004
|
|
5
|
+
http://www.apache.org/licenses/
|
|
6
|
+
|
|
7
|
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
|
8
|
+
|
|
9
|
+
1. Definitions.
|
|
10
|
+
|
|
11
|
+
"License" shall mean the terms and conditions for use, reproduction, and distribution as defined by Sections 1 through 9 of this document.
|
|
12
|
+
|
|
13
|
+
"Licensor" shall mean the copyright owner or entity authorized by the copyright owner that is granting the License.
|
|
14
|
+
|
|
15
|
+
"Legal Entity" shall mean the union of the acting entity and all other entities that control, are controlled by, or are under common control with that entity. For the purposes of this definition, "control" means (i) the power, direct or indirect, to cause the direction or management of such entity, whether by contract or otherwise, or (ii) ownership of fifty percent (50%) or more of the outstanding shares, or (iii) beneficial ownership of such entity.
|
|
16
|
+
|
|
17
|
+
"You" (or "Your") shall mean an individual or Legal Entity exercising permissions granted by this License.
|
|
18
|
+
|
|
19
|
+
"Source" form shall mean the preferred form for making modifications, including but not limited to software source code, documentation source, and configuration files.
|
|
20
|
+
|
|
21
|
+
"Object" form shall mean any form resulting from mechanical transformation or translation of a Source form, including but not limited to compiled object code, generated documentation, and conversions to other media types.
|
|
22
|
+
|
|
23
|
+
"Work" shall mean the work of authorship, whether in Source or Object form, made available under the License, as indicated by a copyright notice that is included in or attached to the work (an example is provided in the Appendix below).
|
|
24
|
+
|
|
25
|
+
"Derivative Works" shall mean any work, whether in Source or Object form, that is based on (or derived from) the Work and for which the editorial revisions, annotations, elaborations, or other modifications represent, as a whole, an original work of authorship. For the purposes of this License, Derivative Works shall not include works that remain separable from, or merely link (or bind by name) to the interfaces of, the Work and Derivative Works thereof.
|
|
26
|
+
|
|
27
|
+
"Contribution" shall mean any work of authorship, including the original version of the Work and any modifications or additions to that Work or Derivative Works thereof, that is intentionally submitted to Licensor for inclusion in the Work by the copyright owner or by an individual or Legal Entity authorized to submit on behalf of the copyright owner. For the purposes of this definition, "submitted" means any form of electronic, verbal, or written communication sent to the Licensor or its representatives, including but not limited to communication on electronic mailing lists, source code control systems, and issue tracking systems that are managed by, or on behalf of, the Licensor for the purpose of discussing and improving the Work, but excluding communication that is conspicuously marked or otherwise designated in writing by the copyright owner as "Not a Contribution."
|
|
28
|
+
|
|
29
|
+
"Contributor" shall mean Licensor and any individual or Legal Entity on behalf of whom a Contribution has been received by Licensor and subsequently incorporated within the Work.
|
|
30
|
+
|
|
31
|
+
2. Grant of Copyright License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable copyright license to reproduce, prepare Derivative Works of, publicly display, publicly perform, sublicense, and distribute the Work and such Derivative Works in Source or Object form.
|
|
32
|
+
|
|
33
|
+
3. Grant of Patent License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer the Work, where such license applies only to those patent claims licensable by such Contributor that are necessarily infringed by their Contribution(s) alone or by combination of their Contribution(s) with the Work to which such Contribution(s) was submitted. If You institute patent litigation against any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Work or a Contribution incorporated within the Work constitutes direct or contributory patent infringement, then any patent licenses granted to You under this License for that Work shall terminate as of the date such litigation is filed.
|
|
34
|
+
|
|
35
|
+
4. Redistribution. You may reproduce and distribute copies of the Work or Derivative Works thereof in any medium, with or without modifications, and in Source or Object form, provided that You meet the following conditions:
|
|
36
|
+
|
|
37
|
+
(a) You must give any other recipients of the Work or Derivative Works a copy of this License; and
|
|
38
|
+
|
|
39
|
+
(b) You must cause any modified files to carry prominent notices stating that You changed the files; and
|
|
40
|
+
|
|
41
|
+
(c) You must retain, in the Source form of any Derivative Works that You distribute, all copyright, patent, trademark, and attribution notices from the Source form of the Work, excluding those notices that do not pertain to any part of the Derivative Works; and
|
|
42
|
+
|
|
43
|
+
(d) If the Work includes a "NOTICE" text file as part of its distribution, then any Derivative Works that You distribute must include a readable copy of the attribution notices contained within such NOTICE file, excluding those notices that do not pertain to any part of the Derivative Works, in at least one of the following places: within a NOTICE text file distributed as part of the Derivative Works; within the Source form or documentation, if provided along with the Derivative Works; or, within a display generated by the Derivative Works, if and wherever such third-party notices normally appear. The contents of the NOTICE file are for informational purposes only and do not modify the License. You may add Your own attribution notices within Derivative Works that You distribute, alongside or as an addendum to the NOTICE text from the Work, provided that such additional attribution notices cannot be construed as modifying the License.
|
|
44
|
+
|
|
45
|
+
You may add Your own copyright statement to Your modifications and may provide additional or different license terms and conditions for use, reproduction, or distribution of Your modifications, or for any such Derivative Works as a whole, provided Your use, reproduction, and distribution of the Work otherwise complies with the conditions stated in this License.
|
|
46
|
+
|
|
47
|
+
5. Submission of Contributions. Unless You explicitly state otherwise, any Contribution intentionally submitted for inclusion in the Work by You to the Licensor shall be under the terms and conditions of this License, without any additional terms or conditions. Notwithstanding the above, nothing herein shall supersede or modify the terms of any separate license agreement you may have executed with Licensor regarding such Contributions.
|
|
48
|
+
|
|
49
|
+
6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor, except as required for reasonable and customary use in describing the origin of the Work and reproducing the content of the NOTICE file.
|
|
50
|
+
|
|
51
|
+
7. Disclaimer of Warranty. Unless required by applicable law or agreed to in writing, Licensor provides the Work (and each Contributor provides its Contributions) on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied, including, without limitation, any warranties or conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A PARTICULAR PURPOSE. You are solely responsible for determining the appropriateness of using or redistributing the Work and assume any risks associated with Your exercise of permissions under this License.
|
|
52
|
+
|
|
53
|
+
8. Limitation of Liability. In no event and under no legal theory, whether in tort (including negligence), contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect, special, incidental, or consequential damages of any character arising as a result of this License or out of the use or inability to use the Work (including but not limited to damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other commercial damages or losses), even if such Contributor has been advised of the possibility of such damages.
|
|
54
|
+
|
|
55
|
+
9. Accepting Warranty or Additional Liability. While redistributing the Work or Derivative Works thereof, You may choose to offer, and charge a fee for, acceptance of support, warranty, indemnity, or other liability obligations and/or rights consistent with this License. However, in accepting such obligations, You may act only on Your own behalf and on Your sole responsibility, not on behalf of any other Contributor, and only if You agree to indemnify, defend, and hold each Contributor harmless for any liability incurred by, or claims asserted against, such Contributor by reason of your accepting any such warranty or additional liability.
|
|
56
|
+
|
|
57
|
+
END OF TERMS AND CONDITIONS
|
|
58
|
+
|
|
59
|
+
APPENDIX: How to apply the Apache License to your work.
|
|
60
|
+
|
|
61
|
+
To apply the Apache License to your work, attach the following boilerplate notice, with the fields enclosed by brackets "[]" replaced with your own identifying information. (Don't include the brackets!) The text should be enclosed in the appropriate comment syntax for the file format. We also recommend that a file or class name and description of purpose be included on the same "printed page" as the copyright notice for easier identification within third-party archives.
|
|
62
|
+
|
|
63
|
+
Copyright [yyyy] [name of copyright owner]
|
|
64
|
+
|
|
65
|
+
Licensed under the Apache License, Version 2.0 (the "License");
|
|
66
|
+
you may not use this file except in compliance with the License.
|
|
67
|
+
You may obtain a copy of the License at
|
|
68
|
+
|
|
69
|
+
http://www.apache.org/licenses/LICENSE-2.0
|
|
70
|
+
|
|
71
|
+
Unless required by applicable law or agreed to in writing, software
|
|
72
|
+
distributed under the License is distributed on an "AS IS" BASIS,
|
|
73
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
74
|
+
See the License for the specific language governing permissions and
|
|
75
|
+
limitations under the License.
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
include incidents.md harnesses/codex.md stay_safe/release.txt
|
stay_safe-0.2.3/PKG-INFO
ADDED
|
@@ -0,0 +1,103 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: stay-safe
|
|
3
|
+
Version: 0.2.3
|
|
4
|
+
Summary: Read Codex configuration files and print a shareable safety report
|
|
5
|
+
License-Expression: Apache-2.0
|
|
6
|
+
Project-URL: Repository, https://github.com/chalkbadger/stay-safe
|
|
7
|
+
Project-URL: Issues, https://github.com/chalkbadger/stay-safe/issues
|
|
8
|
+
Requires-Python: >=3.11
|
|
9
|
+
Description-Content-Type: text/markdown
|
|
10
|
+
License-File: LICENSE
|
|
11
|
+
Dynamic: license-file
|
|
12
|
+
|
|
13
|
+
# Stay safe
|
|
14
|
+
|
|
15
|
+
Check your Codex files for risky settings. See what each warning means and what you can do about it.
|
|
16
|
+
|
|
17
|
+
stay-safe reads settings and saved approvals from files. Only some settings are covered. The report names the defaults it uses and the checks it could not complete. It does not test live connections or what Codex will actually allow. See the [full scope](docs/report.md#full-scope).
|
|
18
|
+
|
|
19
|
+
Here is a finding from a run on a [test configuration](tests/runs_without_a_coding_agent/sample.toml):
|
|
20
|
+
|
|
21
|
+
> **Your config turns off Codex's sandbox.** Commands Codex runs aren't limited to your project, so a mistaken command could damage files elsewhere on your computer.
|
|
22
|
+
>
|
|
23
|
+
> *How it has gone wrong:* Agents have deleted a home directory and wiped a whole drive with one mistaken delete command. [I1](https://github.com/chalkbadger/stay-safe/blob/v0.2.2/incidents.md#i1-home-directory-deleted-by-rm--rf), [I3](https://github.com/chalkbadger/stay-safe/blob/v0.2.2/incidents.md#i3-windows-drive-wiped) (reported, not verified).
|
|
24
|
+
>
|
|
25
|
+
> *What to do:* Set `sandbox_mode = "workspace-write"` (commands can write only in your workspace). Stricter: `sandbox_mode = "read-only"` (commands can't write files).
|
|
26
|
+
>
|
|
27
|
+
> *Change it:* near the top of the file, before any `[section]`, or ask Codex: "Edit `<config-1>` to set `sandbox_mode = "workspace-write"`; show only what you change."
|
|
28
|
+
>
|
|
29
|
+
> *Rule `unrestricted_filesystem` · setting `config-1:sandbox_mode` = `danger-full-access`*
|
|
30
|
+
|
|
31
|
+
It runs on Unix with Python 3.11 or newer. No agent is needed:
|
|
32
|
+
|
|
33
|
+
```sh
|
|
34
|
+
git clone --branch v0.2.2 https://github.com/chalkbadger/stay-safe.git
|
|
35
|
+
cd stay-safe
|
|
36
|
+
python3 -m stay_safe
|
|
37
|
+
```
|
|
38
|
+
|
|
39
|
+
Without Git, [download the v0.2.2 ZIP](https://github.com/chalkbadger/stay-safe/archive/refs/tags/v0.2.2.zip) and extract it. Open a terminal in the folder containing the extracted `stay-safe-0.2.2` folder, then run:
|
|
40
|
+
|
|
41
|
+
```sh
|
|
42
|
+
cd stay-safe-0.2.2
|
|
43
|
+
python3 -m stay_safe
|
|
44
|
+
```
|
|
45
|
+
|
|
46
|
+
By default it reads `~/.codex/config.toml` (`$CODEX_HOME/config.toml` when `CODEX_HOME` is set). [Usage](docs/usage.md) covers the inspected files, options, `--save` and exit status.
|
|
47
|
+
|
|
48
|
+
## Packaged release (publication pending)
|
|
49
|
+
|
|
50
|
+
Version 0.2.3 is prepared for PyPI but is not published yet. Once it is published,
|
|
51
|
+
users with [uv](https://docs.astral.sh/uv/) can run `uvx --python 3.11 stay-safe@0.2.3`.
|
|
52
|
+
The command asks uv for Python 3.11, downloading it if needed; your system Python
|
|
53
|
+
stays unchanged.
|
|
54
|
+
The clone and ZIP commands above are available now. See the
|
|
55
|
+
[publication steps](docs/publishing.md) for the remaining release check.
|
|
56
|
+
|
|
57
|
+
## Ask Codex
|
|
58
|
+
|
|
59
|
+
Paste this into Codex in the ChatGPT app or the CLI:
|
|
60
|
+
|
|
61
|
+
```text
|
|
62
|
+
Download stay-safe at tag v0.2.2 from https://github.com/chalkbadger/stay-safe into
|
|
63
|
+
a new temporary folder and run `python3 -m stay_safe` there. Show the report exactly
|
|
64
|
+
as printed, with its exit status, then your notes under "Agent's own report". Don't
|
|
65
|
+
open auth files or change files outside that folder. If the download fails, stop and
|
|
66
|
+
tell me.
|
|
67
|
+
```
|
|
68
|
+
|
|
69
|
+
In "Ask for approval" mode, approve the one internet request; never choose "don't ask
|
|
70
|
+
again". If Codex can't download, see [Usage](docs/usage.md#if-codex-cant-download).
|
|
71
|
+
|
|
72
|
+
## What it checks and what it can't see
|
|
73
|
+
|
|
74
|
+
- It reads the files listed in [Usage](docs/usage.md) and changes nothing in your setup.
|
|
75
|
+
- Each warning names its rule and the setting behind it, and links to [incident reports](incidents.md). [Reading the report](docs/report.md) explains each part.
|
|
76
|
+
- The [Codex meanings table](harnesses/codex.md) lists the settings it checks and the sources it uses. The report says when it used a default or could not complete a check.
|
|
77
|
+
- The [full scope](docs/report.md#full-scope) explains what it does not inspect.
|
|
78
|
+
- It shows server, app and tool names as written. Check those names before sharing the report.
|
|
79
|
+
|
|
80
|
+
## Challenge a finding
|
|
81
|
+
|
|
82
|
+
Use the [challenge form](https://github.com/chalkbadger/stay-safe/issues/new?template=challenge.yml) to submit the finding, revision, disputed claim and evidence.
|
|
83
|
+
|
|
84
|
+
## Why stay-safe exists
|
|
85
|
+
|
|
86
|
+
stay-safe is part of [chalkbadger](https://github.com/chalkbadger), which works on AI safety through adversarial peer review: every claim a tool makes should be open to challenge, with the evidence to check it. Each finding names the rule and the reported incidents behind it, each claim in [`contracts/claims.md`](contracts/claims.md) names the test that proves it, and anyone can [challenge a finding](#challenge-a-finding). stay-safe is built with [Wirth](https://github.com/chalkbadger/wirth), which keeps those claims and their tests together and reviews every change against them.
|
|
87
|
+
|
|
88
|
+
## Develop
|
|
89
|
+
|
|
90
|
+
Git is required for the development tests. Run `python3 check_tests.py` in a
|
|
91
|
+
development virtual environment. Before the first run, install the build tools
|
|
92
|
+
with `python3 -m pip install build "setuptools>=77" wheel`.
|
|
93
|
+
This checks test-folder package markers before running unittest discovery.
|
|
94
|
+
|
|
95
|
+
## Contribute
|
|
96
|
+
|
|
97
|
+
Share feedback, challenge a finding or propose a change through [CONTRIBUTING.md](CONTRIBUTING.md).
|
|
98
|
+
|
|
99
|
+
## License
|
|
100
|
+
|
|
101
|
+
Licensed under the [Apache License, Version 2.0](LICENSE).
|
|
102
|
+
|
|
103
|
+
References to an *archived* issue or PR point to the private repository that holds this project's earlier working history.
|
|
@@ -0,0 +1,91 @@
|
|
|
1
|
+
# Stay safe
|
|
2
|
+
|
|
3
|
+
Check your Codex files for risky settings. See what each warning means and what you can do about it.
|
|
4
|
+
|
|
5
|
+
stay-safe reads settings and saved approvals from files. Only some settings are covered. The report names the defaults it uses and the checks it could not complete. It does not test live connections or what Codex will actually allow. See the [full scope](docs/report.md#full-scope).
|
|
6
|
+
|
|
7
|
+
Here is a finding from a run on a [test configuration](tests/runs_without_a_coding_agent/sample.toml):
|
|
8
|
+
|
|
9
|
+
> **Your config turns off Codex's sandbox.** Commands Codex runs aren't limited to your project, so a mistaken command could damage files elsewhere on your computer.
|
|
10
|
+
>
|
|
11
|
+
> *How it has gone wrong:* Agents have deleted a home directory and wiped a whole drive with one mistaken delete command. [I1](https://github.com/chalkbadger/stay-safe/blob/v0.2.2/incidents.md#i1-home-directory-deleted-by-rm--rf), [I3](https://github.com/chalkbadger/stay-safe/blob/v0.2.2/incidents.md#i3-windows-drive-wiped) (reported, not verified).
|
|
12
|
+
>
|
|
13
|
+
> *What to do:* Set `sandbox_mode = "workspace-write"` (commands can write only in your workspace). Stricter: `sandbox_mode = "read-only"` (commands can't write files).
|
|
14
|
+
>
|
|
15
|
+
> *Change it:* near the top of the file, before any `[section]`, or ask Codex: "Edit `<config-1>` to set `sandbox_mode = "workspace-write"`; show only what you change."
|
|
16
|
+
>
|
|
17
|
+
> *Rule `unrestricted_filesystem` · setting `config-1:sandbox_mode` = `danger-full-access`*
|
|
18
|
+
|
|
19
|
+
It runs on Unix with Python 3.11 or newer. No agent is needed:
|
|
20
|
+
|
|
21
|
+
```sh
|
|
22
|
+
git clone --branch v0.2.2 https://github.com/chalkbadger/stay-safe.git
|
|
23
|
+
cd stay-safe
|
|
24
|
+
python3 -m stay_safe
|
|
25
|
+
```
|
|
26
|
+
|
|
27
|
+
Without Git, [download the v0.2.2 ZIP](https://github.com/chalkbadger/stay-safe/archive/refs/tags/v0.2.2.zip) and extract it. Open a terminal in the folder containing the extracted `stay-safe-0.2.2` folder, then run:
|
|
28
|
+
|
|
29
|
+
```sh
|
|
30
|
+
cd stay-safe-0.2.2
|
|
31
|
+
python3 -m stay_safe
|
|
32
|
+
```
|
|
33
|
+
|
|
34
|
+
By default it reads `~/.codex/config.toml` (`$CODEX_HOME/config.toml` when `CODEX_HOME` is set). [Usage](docs/usage.md) covers the inspected files, options, `--save` and exit status.
|
|
35
|
+
|
|
36
|
+
## Packaged release (publication pending)
|
|
37
|
+
|
|
38
|
+
Version 0.2.3 is prepared for PyPI but is not published yet. Once it is published,
|
|
39
|
+
users with [uv](https://docs.astral.sh/uv/) can run `uvx --python 3.11 stay-safe@0.2.3`.
|
|
40
|
+
The command asks uv for Python 3.11, downloading it if needed; your system Python
|
|
41
|
+
stays unchanged.
|
|
42
|
+
The clone and ZIP commands above are available now. See the
|
|
43
|
+
[publication steps](docs/publishing.md) for the remaining release check.
|
|
44
|
+
|
|
45
|
+
## Ask Codex
|
|
46
|
+
|
|
47
|
+
Paste this into Codex in the ChatGPT app or the CLI:
|
|
48
|
+
|
|
49
|
+
```text
|
|
50
|
+
Download stay-safe at tag v0.2.2 from https://github.com/chalkbadger/stay-safe into
|
|
51
|
+
a new temporary folder and run `python3 -m stay_safe` there. Show the report exactly
|
|
52
|
+
as printed, with its exit status, then your notes under "Agent's own report". Don't
|
|
53
|
+
open auth files or change files outside that folder. If the download fails, stop and
|
|
54
|
+
tell me.
|
|
55
|
+
```
|
|
56
|
+
|
|
57
|
+
In "Ask for approval" mode, approve the one internet request; never choose "don't ask
|
|
58
|
+
again". If Codex can't download, see [Usage](docs/usage.md#if-codex-cant-download).
|
|
59
|
+
|
|
60
|
+
## What it checks and what it can't see
|
|
61
|
+
|
|
62
|
+
- It reads the files listed in [Usage](docs/usage.md) and changes nothing in your setup.
|
|
63
|
+
- Each warning names its rule and the setting behind it, and links to [incident reports](incidents.md). [Reading the report](docs/report.md) explains each part.
|
|
64
|
+
- The [Codex meanings table](harnesses/codex.md) lists the settings it checks and the sources it uses. The report says when it used a default or could not complete a check.
|
|
65
|
+
- The [full scope](docs/report.md#full-scope) explains what it does not inspect.
|
|
66
|
+
- It shows server, app and tool names as written. Check those names before sharing the report.
|
|
67
|
+
|
|
68
|
+
## Challenge a finding
|
|
69
|
+
|
|
70
|
+
Use the [challenge form](https://github.com/chalkbadger/stay-safe/issues/new?template=challenge.yml) to submit the finding, revision, disputed claim and evidence.
|
|
71
|
+
|
|
72
|
+
## Why stay-safe exists
|
|
73
|
+
|
|
74
|
+
stay-safe is part of [chalkbadger](https://github.com/chalkbadger), which works on AI safety through adversarial peer review: every claim a tool makes should be open to challenge, with the evidence to check it. Each finding names the rule and the reported incidents behind it, each claim in [`contracts/claims.md`](contracts/claims.md) names the test that proves it, and anyone can [challenge a finding](#challenge-a-finding). stay-safe is built with [Wirth](https://github.com/chalkbadger/wirth), which keeps those claims and their tests together and reviews every change against them.
|
|
75
|
+
|
|
76
|
+
## Develop
|
|
77
|
+
|
|
78
|
+
Git is required for the development tests. Run `python3 check_tests.py` in a
|
|
79
|
+
development virtual environment. Before the first run, install the build tools
|
|
80
|
+
with `python3 -m pip install build "setuptools>=77" wheel`.
|
|
81
|
+
This checks test-folder package markers before running unittest discovery.
|
|
82
|
+
|
|
83
|
+
## Contribute
|
|
84
|
+
|
|
85
|
+
Share feedback, challenge a finding or propose a change through [CONTRIBUTING.md](CONTRIBUTING.md).
|
|
86
|
+
|
|
87
|
+
## License
|
|
88
|
+
|
|
89
|
+
Licensed under the [Apache License, Version 2.0](LICENSE).
|
|
90
|
+
|
|
91
|
+
References to an *archived* issue or PR point to the private repository that holds this project's earlier working history.
|
|
@@ -0,0 +1,175 @@
|
|
|
1
|
+
# Codex configuration meanings
|
|
2
|
+
|
|
3
|
+
This is the single interpretation table for the Codex reader. Setting interpretations
|
|
4
|
+
describe a selected configuration-file snapshot or the documented defaults assessed
|
|
5
|
+
under the [no-config-file contract](../contracts/claims.md#works_without_a_config_file-with-no-config-file-the-report-looks-where-codex-looks-and-tells-me-which-codex-defaults-apply). They do not
|
|
6
|
+
establish a session's effective policy, live access, or whether an incident will occur. CLI overrides,
|
|
7
|
+
selected profiles, trusted project layers, managed policies and runtime state are
|
|
8
|
+
uninspected. Missing approval or sandbox selectors are gaps; their defaults are not
|
|
9
|
+
inferred. The [tool-approval] and [shell-environment-input] sources define the
|
|
10
|
+
defaults below. The rules follow the selected file's policy,
|
|
11
|
+
not uninspected runtime layers or actual tool annotations. An open tool catalog can
|
|
12
|
+
retain a parent policy risk; a closed allowlist loses it when every allowed tool is
|
|
13
|
+
denied or has its own approval override. A tool without its own approval mode inherits
|
|
14
|
+
its server's or app's mode, and `apps._default` is a policy template for named apps;
|
|
15
|
+
[each_risk_shows_its_cause](../contracts/claims.md#each_risk_shows_its_cause-for-each-risk-i-see-which-of-my-settings-causes-it-and-real-incidents-where-that-setting-went-wrong)
|
|
16
|
+
says which entry carries each resulting risk. An approval mode alone does not determine a live call’s outcome.
|
|
17
|
+
Tool overrides belong to named apps. Tool allow and deny lists (`enabled_tools`,
|
|
18
|
+
`disabled_tools`) apply only to MCP servers: Codex's [app settings][app-config] have neither.
|
|
19
|
+
Unsupported `apps._default.tools` blocks are skipped with a gap; they create no child tool
|
|
20
|
+
entries or risk facts.
|
|
21
|
+
|
|
22
|
+
For a tool of an enabled app, Codex uses the first that is set: the tool's `enabled`,
|
|
23
|
+
the app's `default_tools_enabled`, then `destructive_enabled` from the app, then
|
|
24
|
+
`_default`, then `true` ([order][app-tool-policy], [tests][app-tool-policy-tests]).
|
|
25
|
+
When `destructive_enabled` decides and is `false`, tools marked destructive, or without a mark, are left out;
|
|
26
|
+
the report names where the value comes from. The rows below call this Codex's order.
|
|
27
|
+
|
|
28
|
+
`documented` means the interpretation is supported by the linked official source;
|
|
29
|
+
`documented in source` identifies behavior read at the cited implementation revision,
|
|
30
|
+
not observed in a live tool call. `documented: retired` and `documented: deprecated`
|
|
31
|
+
identify values the official reference retires or deprecates; both remain unhandled.
|
|
32
|
+
`inferred` identifies stay-safe's stated inference. `open` means the reader does not
|
|
33
|
+
handle that meaning. `confirmed by the maintainer` and `disputed` are available for
|
|
34
|
+
future review; no row currently claims either status.
|
|
35
|
+
|
|
36
|
+
The literal Catalog and Rules clauses are the expected observations and risk IDs
|
|
37
|
+
for the named entry in each independent fixture in
|
|
38
|
+
[meanings.json](../tests/see_every_connection_and_setting/meanings.json). `none` means no entry or no risk,
|
|
39
|
+
respectively. Gap says whether that fixture needs an inspection gap. The
|
|
40
|
+
[table test](../tests/see_every_connection_and_setting/test_meanings.py) compares the reader's result to these
|
|
41
|
+
authored expectations; it does not derive expectations from the reader or its rules.
|
|
42
|
+
Additional positive and negative cases live under
|
|
43
|
+
[each_risk_shows_its_cause](../tests/each_risk_shows_its_cause/test_conditions.py).
|
|
44
|
+
|
|
45
|
+
Codex source links are pinned to OpenAI Codex `rust-v0.167.0`
|
|
46
|
+
(`086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a`) and were re-checked there on 2026-11-09.
|
|
47
|
+
The configuration reference was read on 2026-11-02 for retired values.
|
|
48
|
+
The `auto` default is defined by [AppToolApproval][tool-default].
|
|
49
|
+
|
|
50
|
+
| Setting | Value | Meaning stay-safe uses | Source | Status |
|
|
51
|
+
| --- | --- | --- | --- | --- |
|
|
52
|
+
| `approval_policy` | `never` | Permitted commands do not ask for approval; the sandbox remains separate. Catalog: `never`. Rules: `commands_without_prompts`. Gap: no. | [approval] | documented |
|
|
53
|
+
| `approval_policy` | `on-request` | The model decides when to request approval; this is not blanket preapproval. Catalog: `on-request`. Rules: none. Gap: no. | [approval] | documented |
|
|
54
|
+
| `approval_policy` | `untrusted` | The configuration reference retires this unsupported value; the reader leaves it unhandled. Catalog: none. Rules: none. Gap: yes. | [configuration-reference] | documented: retired |
|
|
55
|
+
| `approval_policy` | `on-failure` | The configuration reference deprecates this value; the reader leaves it unhandled. Catalog: none. Rules: none. Gap: yes. | [configuration-reference] | documented: deprecated |
|
|
56
|
+
| `approval_policy` | `granular` | A category set to `false` rejects its approval requests automatically instead of asking, as `never` does; the reader leaves this policy unhandled. Catalog: none. Rules: none. Gap: yes. | [approval-granular] | open |
|
|
57
|
+
| `sandbox_mode` | `read-only` | The configured legacy sandbox is read-only. Catalog: `read-only`. Rules: none. Gap: no. | [sandbox] | documented |
|
|
58
|
+
| `sandbox_mode` | `workspace-write` | The configured legacy sandbox permits workspace writes. Catalog: `workspace-write`. Rules: none. Gap: no. | [sandbox] | documented |
|
|
59
|
+
| `sandbox_mode` | `danger-full-access` | The configured legacy policy removes sandbox restrictions. Catalog: `danger-full-access`. Rules: `unrestricted_filesystem`. Gap: no. | [sandbox] | documented |
|
|
60
|
+
| `default_permissions` | `selected with sandbox_mode` | Named-profile selection is unhandled and conflicts with the legacy selector; legacy risk facts are suppressed. Catalog: `danger-full-access`, `permission-profile selection is uninspected`. Rules: none. Gap: yes. | [schema] | open |
|
|
61
|
+
| `permissions` | `definitions without selection` | Unselected profile definitions remain unhandled without erasing the explicit legacy sandbox setting. Catalog: `danger-full-access`. Rules: `unrestricted_filesystem`. Gap: yes. | [schema] | open |
|
|
62
|
+
| `sandbox_workspace_write.network_access` | `true` | This workspace-write option controls outbound network access. Catalog: `true`, `workspace-write option`. Rules: `network_access`. Gap: no. | [sandbox] | documented |
|
|
63
|
+
| `sandbox_workspace_write.network_access` | `false` | This workspace-write option controls outbound network access. Catalog: `false`, `workspace-write option`. Rules: none. Gap: no. | [sandbox] | documented |
|
|
64
|
+
| `sandbox_workspace_write.*` | `under read-only` | Workspace-write options are not selected by read-only mode. Catalog: `true`, `inactive or selection unknown`. Rules: none. Gap: no. | [schema] | documented |
|
|
65
|
+
| `mcp_servers.*.enabled` | `false` | A disabled server supplies no enabled program or tool condition. Catalog: `disabled`, `stdio; command withheld`, `tool approval: auto (default)`. Rules: none. Gap: no. | [mcp] | documented |
|
|
66
|
+
| `mcp_servers.*.command` | `nonempty command, no url` | An enabled stdio server starts a configured command; contents are withheld. Catalog: `configured enabled`, `stdio; command withheld`, `tool approval: auto (default)`. Rules: `configured_program`, `self_described_tools`. Gap: no. | [mcp] | documented |
|
|
67
|
+
| `mcp_servers.*.url` | `string, no command` | Codex uses HTTP for any string `url` without `command` ([field][mcp-url], [transport][mcp-transport]); the reader does not check its form, and the endpoint is not evidence of reachability or authentication. Catalog: `configured enabled`, `HTTP; endpoint withheld`, `tool approval: auto (default)`. Rules: `self_described_tools`. Gap: no. | [mcp-transport] | documented in source |
|
|
68
|
+
| `mcp_servers.*.http_headers_helper` | `nonempty command` | An HTTP header helper is a configured program; contents are withheld. Catalog: `configured enabled`, `HTTP; endpoint withheld`, `tool approval: auto (default)`. Rules: `configured_program`, `self_described_tools`. Gap: no. | [mcp] | documented |
|
|
69
|
+
| `mcp_servers.*.enabled_tools` | `named allowlist` | Only listed names are allowed by this file, before the denylist; the list does not establish live availability. Catalog: `configured names: read, files/write`. Rules: none. Gap: no. | [mcp] | documented |
|
|
70
|
+
| `mcp_servers.*.disabled_tools` | `named denylist` | Listed names are removed after applying the allowlist. Catalog: `configured names: delete`. Rules: none. Gap: no. | [mcp] | documented |
|
|
71
|
+
| `mcp_servers.*.default_tools_approval_mode` | `approve` | Preapproval mode. Catalog: `configured enabled`, `HTTP; endpoint withheld`, `tool approval: approve`. Rules: `automatic_tools`. Gap: no. | [tool-approval] | documented in source |
|
|
72
|
+
| `mcp_servers.*.default_tools_approval_mode` | `auto` | Uses server-declared hints for tool approval. Catalog: `configured enabled`, `HTTP; endpoint withheld`, `tool approval: auto`. Rules: `self_described_tools`. Gap: no. | [tool-approval] | documented in source |
|
|
73
|
+
| `mcp_servers.*.default_tools_approval_mode` | `prompt` | Requests approval. Catalog: `configured enabled`, `HTTP; endpoint withheld`, `tool approval: prompt`. Rules: none. Gap: no. | [tool-approval] | documented in source |
|
|
74
|
+
| `mcp_servers.*.default_tools_approval_mode` | `writes` | Uses the server’s read-only hint for tool approval. Catalog: `configured enabled`, `HTTP; endpoint withheld`, `tool approval: writes`. Rules: `self_described_tools`. Gap: no. | [tool-approval] | documented in source |
|
|
75
|
+
| `mcp_servers.*.tools.*.approval_mode` | `approve` | A named tool override replaces the server default when the tool is permitted by its filters. Catalog: `configured enabled`, `tool approval: approve`. Rules: `automatic_tools`. Gap: no. | [mcp] | documented |
|
|
76
|
+
| `mcp_servers.*.tools.*.enabled` | `false` | This app-style field is unsupported for MCP tools and does not disable the configured tool. Catalog: `configured enabled`, `tool approval: approve`. Rules: `automatic_tools`. Gap: yes. | [schema] | open |
|
|
77
|
+
| `apps._default.enabled` | `false, named app without enabled` | Disables no named app: a named app's enablement comes from its own `enabled`, which defaults to `true` ([field][app-enabled-field]), so an app without it stays enabled, as a [recorded Codex 0.167.0 run][app-enabled-run] shows. Catalog: `configured enabled`, `destructive tools enabled by Codex default`, `tool approval: approve`. Rules: `automatic_tools`, `destructive_tools`. Gap: no. | [app-enabled] | documented in source |
|
|
78
|
+
| `apps.*.enabled` | `false` | An explicitly disabled app supplies no enabled tool conditions. Catalog: `disabled`, `tool approval: approve`. Rules: none. Gap: no. | [apps] | documented |
|
|
79
|
+
| `apps.*.default_tools_enabled` | `false` | Codex follows [the order above](#codex-configuration-meanings). This value disables the app's tools without their own `enabled`, whatever `destructive_enabled` says. Catalog: `configured enabled`, `default tools disabled or unknown`, `tool approval: approve`. Rules: none. Gap: no. | [app-tool-policy] | documented in source |
|
|
80
|
+
| `apps.*.default_tools_enabled` | `true` | Codex follows [the order above](#codex-configuration-meanings). This value enables the app's tools without their own `enabled`, destructive or not, whatever `destructive_enabled` says. Catalog: `configured enabled`, `destructive tools enabled by default_tools_enabled`, `tool approval: auto (default)`. Rules: `destructive_tools`, `self_described_tools`. Gap: no. | [app-tool-policy] | documented in source |
|
|
81
|
+
| `apps._default.default_tools_enabled` | `false` | The global default schema has no such field; the reader reports it without inventing inheritance. Catalog: `configured enabled`, `destructive tools enabled by Codex default`, `tool approval: approve`. Rules: `automatic_tools`, `destructive_tools`. Gap: yes. | [schema] | open |
|
|
82
|
+
| `apps.*.default_tools_approval_mode` | `approve inherited from _default` | An app inherits its default tool approval mode unless it overrides it. Catalog: `configured enabled`, `destructive tools enabled by Codex default`, `tool approval: approve`. Rules: `automatic_tools`, `destructive_tools`. Gap: no. | [apps] | documented |
|
|
83
|
+
| `apps.*.enabled_tools` | `on an app` | Codex's app settings have no tool allowlist; the list is a gap and the app keeps its findings. Catalog: `configured enabled`, `destructive tools enabled by Codex default`, `tool approval: approve`. Rules: `automatic_tools`, `destructive_tools`. Gap: yes. | [app-config] | open |
|
|
84
|
+
| `apps.*.disabled_tools` | `on an app` | Codex's app settings have no tool denylist; the list is a gap and a listed tool keeps its findings. Catalog: `configured enabled`, `tool approval: approve`. Rules: `automatic_tools`. Gap: yes. | [app-config] | open |
|
|
85
|
+
| `apps.*.tools.*.approval_mode` | `approve` | A named app tool override replaces its app default. Catalog: `configured enabled`, `tool approval: approve`. Rules: `automatic_tools`. Gap: no. | [apps] | documented |
|
|
86
|
+
| `apps.*.tools.*.enabled` | `false` | A disabled app tool supplies no preapproved-tool condition. Catalog: `disabled or enablement unknown`, `tool approval: approve`. Rules: none. Gap: no. | [apps] | documented |
|
|
87
|
+
| `apps.*.tools.*.enabled` | `true overrides default_tools_enabled` | A tool's own `enabled` comes first in [Codex's order](#codex-configuration-meanings), so this value enables the tool even if it is destructive. Catalog: `configured enabled`, `enabled even if destructive`, `tool approval: approve`. Rules: `automatic_tools`, `destructive_tools`. Gap: no. | [app-tool-policy] | documented in source |
|
|
88
|
+
| `apps.*.destructive_enabled` | `omitted` | Codex follows [the order above](#codex-configuration-meanings). An `_default` table without this field also gives `true` ([default][app-default-destructive]). Catalog: `configured enabled`, `destructive tools enabled by Codex default`, `tool approval: auto (default)`. Rules: `destructive_tools`, `self_described_tools`. Gap: no. | [app-tool-policy] | documented in source |
|
|
89
|
+
| `apps.*.destructive_enabled` | `true inherited from _default` | Codex follows [the order above](#codex-configuration-meanings). Catalog: `configured enabled`, `destructive tools enabled by apps._default.destructive_enabled`, `tool approval: auto (default)`. Rules: `destructive_tools`, `self_described_tools`. Gap: no. | [app-tool-policy] | documented in source |
|
|
90
|
+
| `apps.*.destructive_enabled` | `false inherited from _default` | Codex follows [the order above](#codex-configuration-meanings). Catalog: `configured enabled`, `destructive tools disabled by apps._default.destructive_enabled`, `tool approval: auto (default)`. Rules: `self_described_tools`. Gap: no. | [app-tool-policy] | documented in source |
|
|
91
|
+
| `apps.*.links` | `per-account policy` | Account-specific policy is outside reader coverage and account identities are withheld. Catalog: `configured enabled`, `destructive tools enabled by Codex default`, `tool approval: auto (default)`. Rules: `destructive_tools`, `self_described_tools`. Gap: yes. | [apps] | open |
|
|
92
|
+
| `model_providers.*` | `definition` | The provider name is catalogued; endpoints and credentials are withheld and no provider selection is inferred. Catalog: `configured; endpoint and credentials withheld`. Rules: none. Gap: no. | [schema] | documented |
|
|
93
|
+
| `model_providers.*.auth` | `command-backed authentication` | Command-backed provider authentication is outside reader coverage. Catalog: `configured; endpoint and credentials withheld`. Rules: none. Gap: yes. | [schema] | open |
|
|
94
|
+
| `notify` | `command list with nonempty program` | Notifications may spawn a configured external command; its contents are withheld. Catalog: `notification command configured; command withheld`. Rules: `configured_program`. Gap: no. | [notify-program] | documented in source |
|
|
95
|
+
| `hooks.*` | `command handler with nonempty command` | Codex runs a `type = "command"` handler whose command is not empty, whatever its [matcher][hook-matcher] or [options][hook-handlers] such as `timeout`; trust and enablement in `hooks.state` are uninspected. Catalog: `contents withheld`. Rules: `configured_program`. Gap: yes. | [hooks] | documented in source |
|
|
96
|
+
| `hooks.*` | `no command handler with nonempty command` | Codex skips `prompt` and `agent` handlers ([source][hook-skipped]) and empty commands ([source][hook-empty]); an `mcp_tool` handler calls a tool on an MCP server ([source][hook-mcp-tool]). Catalog: `contents withheld`. Rules: none. Gap: yes. | [hooks] | documented in source |
|
|
97
|
+
| `hooks.state` | `trusts, distrusts or disables a hook` | Codex runs a hook from the user's `config.toml` only when this state enables and trusts it ([gate][hook-trust], [trust check][hook-trust-check], [state source][hook-state]); the reader does not inspect it, so a command handler keeps its risk. Catalog: `contents withheld`. Rules: `configured_program`. Gap: yes. | [hooks] | documented in source |
|
|
98
|
+
| `web_search` | `disabled` | The reader records this recognized configured mode without inferring live network availability. Catalog: `disabled`. Rules: none. Gap: no. | [schema] | documented |
|
|
99
|
+
| `web_search` | `cached` | The reader records this recognized configured mode without inferring live network availability. Catalog: `cached`. Rules: none. Gap: no. | [schema] | documented |
|
|
100
|
+
| `web_search` | `indexed` | The reader records this recognized configured mode without inferring live network availability. Catalog: `indexed`. Rules: none. Gap: no. | [schema] | documented |
|
|
101
|
+
| `web_search` | `live` | The reader records this recognized configured mode without inferring live network availability. Catalog: `live`. Rules: none. Gap: no. | [schema] | documented |
|
|
102
|
+
| `web_search` | `unrecognized string` | An unrecognized mode is withheld and produces a coverage gap. Catalog: none. Rules: none. Gap: yes. | [schema] | open |
|
|
103
|
+
| `mcp_servers.*.default_tools_approval_mode` | `omitted` | The documented auto default applies to an enabled server when no tool override replaces it. Catalog: `configured enabled`, `HTTP; endpoint withheld`, `tool approval: auto (default)`. Rules: `self_described_tools`. Gap: no. | [tool-approval] | documented in source |
|
|
104
|
+
| `apps.*.default_tools_approval_mode` | `omitted without _default mode` | The documented auto default applies to a named enabled app when no app or tool override replaces it. Catalog: `configured enabled`, `destructive tools enabled by Codex default`, `tool approval: auto (default)`. Rules: `destructive_tools`, `self_described_tools`. Gap: no. | [tool-approval] | documented in source |
|
|
105
|
+
| `apps.*.default_tools_approval_mode` | `writes inherited from _default` | The named app inherits the writes policy from its default; the hint-dependent meaning is defined in the mode rows above. Catalog: `configured enabled`, `destructive tools enabled by Codex default`, `tool approval: writes`. Rules: `destructive_tools`, `self_described_tools`. Gap: no. | [apps] | documented |
|
|
106
|
+
| `shell_environment_policy` | `omitted` | Catalog: `inherit: all (default)`, `ignore_default_excludes: true (default)`. Rules: `environment_inheritance`. Gap: no. | [shell-environment-input] | documented in source |
|
|
107
|
+
| `shell_environment_policy.inherit` | `all` | Catalog: `inherit: all (explicit)`, `ignore_default_excludes: true (default)`. Rules: `environment_inheritance`. Gap: no. | [shell-environment] | documented in source |
|
|
108
|
+
| `shell_environment_policy.inherit` | `core` | Catalog: `inherit: core (explicit)`, `ignore_default_excludes: true (default)`. Rules: none. Gap: no. | [shell-environment] | documented in source |
|
|
109
|
+
| `shell_environment_policy.inherit` | `none` | Catalog: `inherit: none (explicit)`, `ignore_default_excludes: true (default)`. Rules: none. Gap: no. | [shell-environment] | documented in source |
|
|
110
|
+
| `shell_environment_policy.ignore_default_excludes` | `true` | Catalog: `inherit: all (default)`, `ignore_default_excludes: true (explicit)`. Rules: `environment_inheritance`. Gap: no. | [shell-environment] | documented in source |
|
|
111
|
+
| `shell_environment_policy.ignore_default_excludes` | `false` | Catalog: `inherit: all (default)`, `ignore_default_excludes: false (explicit)`. Rules: `environment_inheritance`. Gap: no. | [shell-environment] | documented in source |
|
|
112
|
+
| `shell_environment_policy` | `additional fields` | Catalog: none. Rules: none. Gap: yes. | [shell-environment] | open |
|
|
113
|
+
| `features.apps` | `omitted` | Codex's apps feature is on by default ([default][apps-feature]). When Codex is signed in with ChatGPT, it adds an MCP connection to ChatGPT named `codex_apps` ([connection][apps-connection], [sign-in][apps-sign-in]); `false` removes it. The reader does not check sign-in. Catalog: `ChatGPT Apps`, `exists only if Codex is signed in with ChatGPT, which is not checked`, `features.apps = false turns it off`. Rules: none. Gap: no. | [apps-feature] | documented in source |
|
|
114
|
+
|
|
115
|
+
## Saved approval records
|
|
116
|
+
|
|
117
|
+
The [Codex writer][saved-writer] appends records such as
|
|
118
|
+
`prefix_rule(pattern=["curl"], decision="allow")` to the
|
|
119
|
+
[default rules file][saved-runtime]. This reader covers that one-line layout,
|
|
120
|
+
with a nonempty JSON string array and an explicit `allow` decision. Other
|
|
121
|
+
layouts and `network_rule` records remain outside coverage, as do `prompt` and
|
|
122
|
+
`forbidden` records, whose stricter decision could override a listed `allow`.
|
|
123
|
+
It inventories matching lines, not the structure of a Starlark program. A record covers every command whose
|
|
124
|
+
first words equal its pattern ([matching][saved-match]), so a pattern with more words covers
|
|
125
|
+
fewer commands.
|
|
126
|
+
|
|
127
|
+
A record's words are withheld, except a first word on the program list in the
|
|
128
|
+
[saved-approval contract](../contracts/claims.md#lists_commands_codex_stops_asking_about-the-report-lists-the-commands-i-told-codex-to-stop-asking-about-and-tells-me-when-it-couldnt-read-or-check-a-rules-file), which is shown as
|
|
129
|
+
`program: <name>`. The list changes only by pull request.
|
|
130
|
+
|
|
131
|
+
The [saved-approval contract](../contracts/claims.md#lists_commands_codex_stops_asking_about-the-report-lists-the-commands-i-told-codex-to-stop-asking-about-and-tells-me-when-it-couldnt-read-or-check-a-rules-file) defines how the
|
|
132
|
+
report treats records inside and outside coverage; [share_without_exposing_secrets](../contracts/claims.md#share_without_exposing_secrets-i-can-share-the-report-without-exposing-my-commands-addresses-tokens-or-environment-values-names-appear-as-i-wrote-them-and-the-report-reminds-me-to-check-them-first)
|
|
133
|
+
covers disclosure. The broad label is stay-safe's inference, not a Codex score.
|
|
134
|
+
[Decision ordering][saved-decisions], [matching precedence][saved-precedence]
|
|
135
|
+
and [runtime sandbox handling][saved-runtime] establish the declared meaning.
|
|
136
|
+
|
|
137
|
+
[saved-writer]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/execpolicy/src/amend.rs
|
|
138
|
+
[saved-runtime]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/core/src/exec_policy.rs
|
|
139
|
+
[saved-match]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/execpolicy/src/rule.rs#L55-L72
|
|
140
|
+
[saved-precedence]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/execpolicy/src/policy.rs#L584-L595
|
|
141
|
+
[saved-decisions]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/execpolicy/src/decision.rs
|
|
142
|
+
|
|
143
|
+
[schema]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/core/config.schema.json
|
|
144
|
+
[approval]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/protocol/src/protocol.rs
|
|
145
|
+
[approval-granular]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/protocol/src/protocol.rs#L1158-L1168
|
|
146
|
+
[sandbox]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/protocol/src/protocol.rs
|
|
147
|
+
[mcp]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/config/src/mcp_types.rs
|
|
148
|
+
[tool-approval]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/core/src/mcp_tool_call.rs#L2597-L2627
|
|
149
|
+
[apps]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/config/src/types.rs
|
|
150
|
+
[app-tool-policy]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/connectors/src/app_tool_policy.rs#L279-L305
|
|
151
|
+
[app-config]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/config/src/types.rs#L534-L576
|
|
152
|
+
[app-default-destructive]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/config/src/types.rs#L471-L476
|
|
153
|
+
[app-enabled]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/connectors/src/app_tool_policy.rs#L174-L185
|
|
154
|
+
[app-enabled-field]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/config/src/types.rs#L537-L540
|
|
155
|
+
[apps-feature]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/features/src/lib.rs#L1658-L1663
|
|
156
|
+
[apps-connection]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/core/src/mcp.rs#L464-L479
|
|
157
|
+
[apps-sign-in]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/codex-mcp/src/mcp/mod.rs#L358-L360
|
|
158
|
+
[app-enabled-run]: https://github.com/chalkbadger/stay-safe-archive/issues/90#issuecomment-6726050010
|
|
159
|
+
[app-tool-policy-tests]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/connectors/src/app_tool_policy_tests.rs#L685-L770
|
|
160
|
+
[hooks]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/hooks/src/engine/discovery.rs
|
|
161
|
+
[mcp-url]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/config/src/mcp_types.rs#L407
|
|
162
|
+
[mcp-transport]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/config/src/mcp_types.rs#L512-L566
|
|
163
|
+
[hook-matcher]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/hooks/src/engine/matcher.rs#L32-L40
|
|
164
|
+
[hook-handlers]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/protocol/src/hooks.rs#L140-L188
|
|
165
|
+
[hook-skipped]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/hooks/src/engine/discovery.rs#L782-L801
|
|
166
|
+
[hook-empty]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/hooks/src/engine/discovery.rs#L665-L671
|
|
167
|
+
[hook-mcp-tool]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/hooks/src/engine/discovery.rs#L729-L781
|
|
168
|
+
[hook-trust]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/hooks/src/engine/discovery.rs#L917-L923
|
|
169
|
+
[hook-trust-check]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/hooks/src/engine/discovery.rs#L1000-L1027
|
|
170
|
+
[hook-state]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/hooks/src/config_rules.rs#L15-L45
|
|
171
|
+
[notify-program]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/hooks/src/registry.rs#L407-L421
|
|
172
|
+
[configuration-reference]: https://learn.chatgpt.com/docs/config-file/config-reference
|
|
173
|
+
[tool-default]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/config/src/mcp_types.rs#L26-L34
|
|
174
|
+
[shell-environment]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/protocol/src/config_types.rs#L205-L272
|
|
175
|
+
[shell-environment-input]: https://github.com/openai/codex/blob/086ccc833bc9cf01024ad2dd6af7dbd421b6bf9a/codex-rs/config/src/shell_environment_policy.rs#L133-L169
|