sovereign-shield-ch 0.1.0__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.
@@ -0,0 +1,45 @@
1
+ # --- secrets ---
2
+ .env
3
+ .env.*
4
+ !.env.example
5
+ *.pem
6
+ *.key
7
+
8
+ # --- python ---
9
+ __pycache__/
10
+ *.py[cod]
11
+ *$py.class
12
+ *.egg-info/
13
+ *.egg
14
+ build/
15
+ dist/
16
+ .Python
17
+ .venv/
18
+ venv/
19
+ .python-version
20
+
21
+ # --- python tooling caches ---
22
+ .ruff_cache/
23
+ .mypy_cache/
24
+ .pytest_cache/
25
+ .coverage
26
+ .coverage.*
27
+ htmlcov/
28
+ coverage.xml
29
+
30
+ # --- node / next.js (web/ demo) ---
31
+ node_modules/
32
+ .next/
33
+ out/
34
+ .vercel/
35
+ *.tsbuildinfo
36
+ next-env.d.ts
37
+
38
+ # --- editor / OS / agent ---
39
+ .vscode/
40
+ .idea/
41
+ .claude/
42
+ .DS_Store
43
+ *.swp
44
+ *.swo
45
+ *~
@@ -0,0 +1,201 @@
1
+ Apache License
2
+ Version 2.0, January 2004
3
+ http://www.apache.org/licenses/
4
+
5
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
6
+
7
+ 1. Definitions.
8
+
9
+ "License" shall mean the terms and conditions for use, reproduction,
10
+ and distribution as defined by Sections 1 through 9 of this document.
11
+
12
+ "Licensor" shall mean the copyright owner or entity authorized by
13
+ the copyright owner that is granting the License.
14
+
15
+ "Legal Entity" shall mean the union of the acting entity and all
16
+ other entities that control, are controlled by, or are under common
17
+ control with that entity. For the purposes of this definition,
18
+ "control" means (i) the power, direct or indirect, to cause the
19
+ direction or management of such entity, whether by contract or
20
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
21
+ outstanding shares, or (iii) beneficial ownership of such entity.
22
+
23
+ "You" (or "Your") shall mean an individual or Legal Entity
24
+ exercising permissions granted by this License.
25
+
26
+ "Source" form shall mean the preferred form for making modifications,
27
+ including but not limited to software source code, documentation
28
+ source, and configuration files.
29
+
30
+ "Object" form shall mean any form resulting from mechanical
31
+ transformation or translation of a Source form, including but
32
+ not limited to compiled object code, generated documentation,
33
+ and conversions to other media types.
34
+
35
+ "Work" shall mean the work of authorship, whether in Source or
36
+ Object form, made available under the License, as indicated by a
37
+ copyright notice that is included in or attached to the work
38
+ (an example is provided in the Appendix below).
39
+
40
+ "Derivative Works" shall mean any work, whether in Source or Object
41
+ form, that is based on (or derived from) the Work and for which the
42
+ editorial revisions, annotations, elaborations, or other modifications
43
+ represent, as a whole, an original work of authorship. For the purposes
44
+ of this License, Derivative Works shall not include works that remain
45
+ separable from, or merely link (or bind by name) to the interfaces of,
46
+ the Work and Derivative Works thereof.
47
+
48
+ "Contribution" shall mean any work of authorship, including
49
+ the original version of the Work and any modifications or additions
50
+ to that Work or Derivative Works thereof, that is intentionally
51
+ submitted to Licensor for inclusion in the Work by the copyright owner
52
+ or by an individual or Legal Entity authorized to submit on behalf of
53
+ the copyright owner. For the purposes of this definition, "submitted"
54
+ means any form of electronic, verbal, or written communication sent
55
+ to the Licensor or its representatives, including but not limited to
56
+ communication on electronic mailing lists, source code control systems,
57
+ and issue tracking systems that are managed by, or on behalf of, the
58
+ Licensor for the purpose of discussing and improving the Work, but
59
+ excluding communication that is conspicuously marked or otherwise
60
+ designated in writing by the copyright owner as "Not a Contribution."
61
+
62
+ "Contributor" shall mean Licensor and any individual or Legal Entity
63
+ on behalf of whom a Contribution has been received by Licensor and
64
+ subsequently incorporated within the Work.
65
+
66
+ 2. Grant of Copyright License. Subject to the terms and conditions of
67
+ this License, each Contributor hereby grants to You a perpetual,
68
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
69
+ copyright license to reproduce, prepare Derivative Works of,
70
+ publicly display, publicly perform, sublicense, and distribute the
71
+ Work and such Derivative Works in Source or Object form.
72
+
73
+ 3. Grant of Patent License. Subject to the terms and conditions of
74
+ this License, each Contributor hereby grants to You a perpetual,
75
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
76
+ (except as stated in this section) patent license to make, have made,
77
+ use, offer to sell, sell, import, and otherwise transfer the Work,
78
+ where such license applies only to those patent claims licensable
79
+ by such Contributor that are necessarily infringed by their
80
+ Contribution(s) alone or by combination of their Contribution(s)
81
+ with the Work to which such Contribution(s) was submitted. If You
82
+ institute patent litigation against any entity (including a
83
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
84
+ or a Contribution incorporated within the Work constitutes direct
85
+ or contributory patent infringement, then any patent licenses
86
+ granted to You under this License for that Work shall terminate
87
+ as of the date such litigation is filed.
88
+
89
+ 4. Redistribution. You may reproduce and distribute copies of the
90
+ Work or Derivative Works thereof in any medium, with or without
91
+ modifications, and in Source or Object form, provided that You
92
+ meet the following conditions:
93
+
94
+ (a) You must give any other recipients of the Work or
95
+ Derivative Works a copy of this License; and
96
+
97
+ (b) You must cause any modified files to carry prominent notices
98
+ stating that You changed the files; and
99
+
100
+ (c) You must retain, in the Source form of any Derivative Works
101
+ that You distribute, all copyright, patent, trademark, and
102
+ attribution notices from the Source form of the Work,
103
+ excluding those notices that do not pertain to any part of
104
+ the Derivative Works; and
105
+
106
+ (d) If the Work includes a "NOTICE" text file as part of its
107
+ distribution, then any Derivative Works that You distribute must
108
+ include a readable copy of the attribution notices contained
109
+ within such NOTICE file, excluding those notices that do not
110
+ pertain to any part of the Derivative Works, in at least one
111
+ of the following places: within a NOTICE text file distributed
112
+ as part of the Derivative Works; within the Source form or
113
+ documentation, if provided along with the Derivative Works; or,
114
+ within a display generated by the Derivative Works, if and
115
+ wherever such third-party notices normally appear. The contents
116
+ of the NOTICE file are for informational purposes only and
117
+ do not modify the License. You may add Your own attribution
118
+ notices within Derivative Works that You distribute, alongside
119
+ or as an addendum to the NOTICE text from the Work, provided
120
+ that such additional attribution notices cannot be construed
121
+ as modifying the License.
122
+
123
+ You may add Your own copyright statement to Your modifications and
124
+ may provide additional or different license terms and conditions
125
+ for use, reproduction, or distribution of Your modifications, or
126
+ for any such Derivative Works as a whole, provided Your use,
127
+ reproduction, and distribution of the Work otherwise complies with
128
+ the conditions stated in this License.
129
+
130
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
131
+ any Contribution intentionally submitted for inclusion in the Work
132
+ by You to the Licensor shall be under the terms and conditions of
133
+ this License, without any additional terms or conditions.
134
+ Notwithstanding the above, nothing herein shall supersede or modify
135
+ the terms of any separate license agreement you may have executed
136
+ with Licensor regarding such Contributions.
137
+
138
+ 6. Trademarks. This License does not grant permission to use the trade
139
+ names, trademarks, service marks, or product names of the Licensor,
140
+ except as required for reasonable and customary use in describing the
141
+ origin of the Work and reproducing the content of the NOTICE file.
142
+
143
+ 7. Disclaimer of Warranty. Unless required by applicable law or
144
+ agreed to in writing, Licensor provides the Work (and each
145
+ Contributor provides its Contributions) on an "AS IS" BASIS,
146
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
147
+ implied, including, without limitation, any warranties or conditions
148
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
149
+ PARTICULAR PURPOSE. You are solely responsible for determining the
150
+ appropriateness of using or redistributing the Work and assume any
151
+ risks associated with Your exercise of permissions under this License.
152
+
153
+ 8. Limitation of Liability. In no event and under no legal theory,
154
+ whether in tort (including negligence), contract, or otherwise,
155
+ unless required by applicable law (such as deliberate and grossly
156
+ negligent acts) or agreed to in writing, shall any Contributor be
157
+ liable to You for damages, including any direct, indirect, special,
158
+ incidental, or consequential damages of any character arising as a
159
+ result of this License or out of the use or inability to use the
160
+ Work (including but not limited to damages for loss of goodwill,
161
+ work stoppage, computer failure or malfunction, or any and all
162
+ other commercial damages or losses), even if such Contributor
163
+ has been advised of the possibility of such damages.
164
+
165
+ 9. Accepting Warranty or Additional Liability. While redistributing
166
+ the Work or Derivative Works thereof, You may choose to offer,
167
+ and charge a fee for, acceptance of support, warranty, indemnity,
168
+ or other liability obligations and/or rights consistent with this
169
+ License. However, in accepting such obligations, You may act only
170
+ on Your own behalf and on Your sole responsibility, not on behalf
171
+ of any other Contributor, and only if You agree to indemnify,
172
+ defend, and hold each Contributor harmless for any liability
173
+ incurred by, or claims asserted against, such Contributor by reason
174
+ of your accepting any such warranty or additional liability.
175
+
176
+ END OF TERMS AND CONDITIONS
177
+
178
+ APPENDIX: How to apply the Apache License to your work.
179
+
180
+ To apply the Apache License to your work, attach the following
181
+ boilerplate notice, with the fields enclosed by brackets "[]"
182
+ replaced with your own identifying information. (Don't include
183
+ the brackets!) The text should be enclosed in the appropriate
184
+ comment syntax for the file format. We also recommend that a
185
+ file or class name and description of purpose be included on the
186
+ same "printed page" as the copyright notice for easier
187
+ identification within third-party archives.
188
+
189
+ Copyright [yyyy] [name of copyright owner]
190
+
191
+ Licensed under the Apache License, Version 2.0 (the "License");
192
+ you may not use this file except in compliance with the License.
193
+ You may obtain a copy of the License at
194
+
195
+ http://www.apache.org/licenses/LICENSE-2.0
196
+
197
+ Unless required by applicable law or agreed to in writing, software
198
+ distributed under the License is distributed on an "AS IS" BASIS,
199
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
200
+ See the License for the specific language governing permissions and
201
+ limitations under the License.
@@ -0,0 +1,16 @@
1
+ Sovereign Shield
2
+ Copyright 2026 acoseac
3
+
4
+ This product includes software developed by acoseac and contributors
5
+ under the Apache License, Version 2.0.
6
+
7
+ Sovereign Shield is an engineering utility for programmatic privacy
8
+ mitigation (deterministic Swiss/EU PII detection and reversible
9
+ tokenization). It is NOT an automated guarantee of regulatory compliance
10
+ under the Swiss Federal Act on Data Protection (FADP) or the EU GDPR, and
11
+ it is not legal advice. Structural, context-independent detection cannot
12
+ catch every leak vector. Use it alongside audit logs, a DPIA where
13
+ required, and human review. See the README for the full disclaimer.
14
+
15
+ The deterministic detection core is extracted from the K.E.V.I.N.
16
+ adversarial-testing research project.
@@ -0,0 +1,159 @@
1
+ Metadata-Version: 2.4
2
+ Name: sovereign-shield-ch
3
+ Version: 0.1.0
4
+ Summary: Deterministic, offline Swiss/EU (FADP/GDPR) PII gateway: tokenize personal data before a cloud LLM sees it, restore it on the way back.
5
+ Project-URL: Homepage, https://github.com/acoseac/sovereign-shield
6
+ Project-URL: Repository, https://github.com/acoseac/sovereign-shield
7
+ Project-URL: Issues, https://github.com/acoseac/sovereign-shield/issues
8
+ Project-URL: Live demo, https://shield.ars.md
9
+ Author: acoseac
10
+ License: Apache-2.0
11
+ License-File: LICENSE
12
+ License-File: NOTICE
13
+ Keywords: anonymization,compliance,data-sovereignty,fadp,gdpr,llm,pii,privacy,swiss,tokenization
14
+ Classifier: Development Status :: 4 - Beta
15
+ Classifier: Intended Audience :: Developers
16
+ Classifier: License :: OSI Approved :: Apache Software License
17
+ Classifier: Programming Language :: Python :: 3.12
18
+ Classifier: Topic :: Security
19
+ Classifier: Topic :: Software Development :: Libraries
20
+ Classifier: Typing :: Typed
21
+ Requires-Python: >=3.12
22
+ Provides-Extra: dev
23
+ Requires-Dist: mypy>=1.10; extra == 'dev'
24
+ Requires-Dist: pytest>=8.0; extra == 'dev'
25
+ Requires-Dist: ruff>=0.6; extra == 'dev'
26
+ Provides-Extra: gateway
27
+ Requires-Dist: langchain-core>=0.3; extra == 'gateway'
28
+ Requires-Dist: langchain-google-genai>=2.0; extra == 'gateway'
29
+ Description-Content-Type: text/markdown
30
+
31
+ # Sovereign Shield
32
+
33
+ **Use any cloud LLM. Keep the personal data in Switzerland.**
34
+
35
+ A deterministic, offline gateway for Swiss/EU (FADP / GDPR) personal data. It
36
+ tokenizes structured identifiers **locally** — turning `756.1234.5678.97` into
37
+ `[AHV_1]` — so a prompt can go to Gemini, Claude, or any model without a real
38
+ identifier ever crossing the border, then restores the real values in the reply
39
+ on the way back. Detection is regex + checksum, not ML: zero dependencies, zero
40
+ latency, and it **cannot be talked out of a match**.
41
+
42
+ > ⚠️ **Disclaimer.** Sovereign Shield is an engineering utility that aids
43
+ > programmatic privacy mitigation. It is **not** an automated guarantee of
44
+ > regulatory compliance under the Swiss Federal Act on Data Protection (FADP) or
45
+ > the EU GDPR, and it is **not legal advice**. Context-dependent leak vectors
46
+ > (free-text names, encoded data, semantics) can still slip past a structural,
47
+ > deterministic layer. Use it alongside a DPIA where required, audit logs, and
48
+ > human review — as the outer, deliberately-dumb layer of a defence-in-depth stack.
49
+
50
+ [**See it live → shield.ars.md**](https://shield.ars.md) · deterministic, in-browser, no API key.
51
+
52
+ ## Install
53
+
54
+ ```bash
55
+ pip install sovereign-shield-ch # core: stdlib-only, zero dependencies
56
+ pip install "sovereign-shield-ch[gateway]" # + the optional LangChain proxy
57
+ ```
58
+
59
+ Requires Python 3.12+.
60
+
61
+ ## Quickstart
62
+
63
+ ```python
64
+ from sovereign_shield import SovereignShield
65
+
66
+ shield = SovereignShield()
67
+
68
+ raw = ("Guten Tag. Meine AHV-Nummer ist 756.1234.5678.97. Bitte die Praemie auf "
69
+ "IBAN CH9300762011623852957 zurueckerstatten. Erreichbar unter "
70
+ "+41 79 214 88 03 oder hans.muster@bluewin.ch.")
71
+
72
+ # 1. De-identify locally. `safe` is all that crosses the border.
73
+ safe, ctx = shield.sanitize(raw)
74
+ # safe -> "Guten Tag. Meine AHV-Nummer ist [AHV_1]. Bitte die Praemie auf
75
+ # IBAN [IBAN_1] zurueckerstatten. Erreichbar unter [PHONE_1] oder [EMAIL_1]."
76
+ print(ctx.audit()) # {'ch_ahv': 1, 'iban': 1, 'ch_phone': 1, 'email': 1}
77
+
78
+ # 2. Call any cloud LLM on the placeholders (it never sees a real value).
79
+ answer = call_your_llm(safe)
80
+
81
+ # 3. Restore the real values locally before serving the user.
82
+ result = shield.rehydrate(answer, ctx)
83
+ print(result.text) # real AHV / IBAN / phone / email swapped back in
84
+ print(result.clean) # True if the model didn't mangle a placeholder
85
+ ```
86
+
87
+ `sanitize` is **fail-closed**: if any structured identifier would survive into
88
+ `safe`, it raises `DataLeakError` instead of leaking. `rehydrate` is strict and
89
+ deterministic, and reports any placeholder the model mangled or invented
90
+ (`result.leftover`) so you never ship a broken `[AHV_1` to a user.
91
+
92
+ ### Transparent LangChain proxy
93
+
94
+ With the `[gateway]` extra, wrap any LangChain chat model and call it as usual —
95
+ sanitize-out and rehydrate-in happen under the hood:
96
+
97
+ ```python
98
+ from langchain_google_genai import ChatGoogleGenerativeAI
99
+ from sovereign_shield.gateway import ShieldedChatModel
100
+
101
+ llm = ShieldedChatModel(ChatGoogleGenerativeAI(model="gemini-2.5-flash", temperature=0.3))
102
+
103
+ reply = llm.invoke("Refund AHV 756.1234.5678.97 to IBAN CH9300762011623852957.")
104
+ print(reply.content) # real values restored
105
+ print(reply.additional_kwargs["sovereign_shield"]) # {'kept_on_shore': 2, 'leftover': []}
106
+ ```
107
+
108
+ ## What it detects
109
+
110
+ Deterministic *shape regex + checksum* — the checksum rejects look-alikes so the
111
+ guard never trips on a random 13-digit string. Separators are stripped first, so
112
+ `756.1234.5678.97` and `756 1234 5678 97` validate identically.
113
+
114
+ | Category | Identifier | Validation |
115
+ |---|---|---|
116
+ | `ch_ahv` | Swiss AHV / AVS number | EAN-13 check digit |
117
+ | `iban` | CH / LI IBAN | ISO-7064 mod-97 |
118
+ | `credit_card` | Card PAN | Luhn |
119
+ | `ch_phone` | Swiss phone | shape only |
120
+ | `email` | Email | shape only |
121
+ | `dob` | Date of birth | off by default (bare dates false-positive) |
122
+
123
+ **Scope: structured identifiers only.** Person names and street addresses are
124
+ *not* detected — they need an NER model, which would forfeit the deterministic,
125
+ zero-dependency guarantee. Plug your own via `SovereignShield(extra_detectors=[...])`
126
+ (see `SpanDetector`); overlapping spans are dropped fail-closed.
127
+
128
+ **Not encoding-robust.** A model that base64s or ciphers an identifier defeats the
129
+ regex. Separator/whitespace reformatting is handled; encoding is not.
130
+
131
+ ## How it works
132
+
133
+ The thesis, proven in the [K.E.V.I.N.](https://github.com/acoseac/kevin) red-team
134
+ research this is extracted from: you can't close a data leak from *inside* the
135
+ model — a jailbreak, a pretext, or a forced output schema will make it disclose.
136
+ So you put a **deterministic, offline boundary** around the model instead. Sovereign
137
+ Shield is that boundary, as a library: detect → tokenize → (model) → restore.
138
+
139
+ The browser demo ships a TypeScript port of the exact same detectors, kept
140
+ **byte-for-byte in parity** with this Python source by a generated vector suite —
141
+ so redaction on the client and on the server can never silently drift.
142
+
143
+ ## Development
144
+
145
+ ```bash
146
+ pip install -e ".[dev]"
147
+ pytest # unit + round-trip suite
148
+ ruff check . && ruff format --check . && mypy
149
+ python scripts/gen_shield_vectors.py --check # Python parity vectors current
150
+ cd web && npm install && npm run parity # TS shield reproduces them exactly
151
+ ```
152
+
153
+ The `web/` directory is the live demo (Next.js). See [web/README.md](web/README.md).
154
+
155
+ ## Credits & license
156
+
157
+ Extracted from the [K.E.V.I.N.](https://github.com/acoseac/kevin) adversarial-testing
158
+ project; background in the [FADP AI-gateway write-up](https://www.ars.md/blog).
159
+ Licensed under [Apache-2.0](LICENSE).
@@ -0,0 +1,129 @@
1
+ # Sovereign Shield
2
+
3
+ **Use any cloud LLM. Keep the personal data in Switzerland.**
4
+
5
+ A deterministic, offline gateway for Swiss/EU (FADP / GDPR) personal data. It
6
+ tokenizes structured identifiers **locally** — turning `756.1234.5678.97` into
7
+ `[AHV_1]` — so a prompt can go to Gemini, Claude, or any model without a real
8
+ identifier ever crossing the border, then restores the real values in the reply
9
+ on the way back. Detection is regex + checksum, not ML: zero dependencies, zero
10
+ latency, and it **cannot be talked out of a match**.
11
+
12
+ > ⚠️ **Disclaimer.** Sovereign Shield is an engineering utility that aids
13
+ > programmatic privacy mitigation. It is **not** an automated guarantee of
14
+ > regulatory compliance under the Swiss Federal Act on Data Protection (FADP) or
15
+ > the EU GDPR, and it is **not legal advice**. Context-dependent leak vectors
16
+ > (free-text names, encoded data, semantics) can still slip past a structural,
17
+ > deterministic layer. Use it alongside a DPIA where required, audit logs, and
18
+ > human review — as the outer, deliberately-dumb layer of a defence-in-depth stack.
19
+
20
+ [**See it live → shield.ars.md**](https://shield.ars.md) · deterministic, in-browser, no API key.
21
+
22
+ ## Install
23
+
24
+ ```bash
25
+ pip install sovereign-shield-ch # core: stdlib-only, zero dependencies
26
+ pip install "sovereign-shield-ch[gateway]" # + the optional LangChain proxy
27
+ ```
28
+
29
+ Requires Python 3.12+.
30
+
31
+ ## Quickstart
32
+
33
+ ```python
34
+ from sovereign_shield import SovereignShield
35
+
36
+ shield = SovereignShield()
37
+
38
+ raw = ("Guten Tag. Meine AHV-Nummer ist 756.1234.5678.97. Bitte die Praemie auf "
39
+ "IBAN CH9300762011623852957 zurueckerstatten. Erreichbar unter "
40
+ "+41 79 214 88 03 oder hans.muster@bluewin.ch.")
41
+
42
+ # 1. De-identify locally. `safe` is all that crosses the border.
43
+ safe, ctx = shield.sanitize(raw)
44
+ # safe -> "Guten Tag. Meine AHV-Nummer ist [AHV_1]. Bitte die Praemie auf
45
+ # IBAN [IBAN_1] zurueckerstatten. Erreichbar unter [PHONE_1] oder [EMAIL_1]."
46
+ print(ctx.audit()) # {'ch_ahv': 1, 'iban': 1, 'ch_phone': 1, 'email': 1}
47
+
48
+ # 2. Call any cloud LLM on the placeholders (it never sees a real value).
49
+ answer = call_your_llm(safe)
50
+
51
+ # 3. Restore the real values locally before serving the user.
52
+ result = shield.rehydrate(answer, ctx)
53
+ print(result.text) # real AHV / IBAN / phone / email swapped back in
54
+ print(result.clean) # True if the model didn't mangle a placeholder
55
+ ```
56
+
57
+ `sanitize` is **fail-closed**: if any structured identifier would survive into
58
+ `safe`, it raises `DataLeakError` instead of leaking. `rehydrate` is strict and
59
+ deterministic, and reports any placeholder the model mangled or invented
60
+ (`result.leftover`) so you never ship a broken `[AHV_1` to a user.
61
+
62
+ ### Transparent LangChain proxy
63
+
64
+ With the `[gateway]` extra, wrap any LangChain chat model and call it as usual —
65
+ sanitize-out and rehydrate-in happen under the hood:
66
+
67
+ ```python
68
+ from langchain_google_genai import ChatGoogleGenerativeAI
69
+ from sovereign_shield.gateway import ShieldedChatModel
70
+
71
+ llm = ShieldedChatModel(ChatGoogleGenerativeAI(model="gemini-2.5-flash", temperature=0.3))
72
+
73
+ reply = llm.invoke("Refund AHV 756.1234.5678.97 to IBAN CH9300762011623852957.")
74
+ print(reply.content) # real values restored
75
+ print(reply.additional_kwargs["sovereign_shield"]) # {'kept_on_shore': 2, 'leftover': []}
76
+ ```
77
+
78
+ ## What it detects
79
+
80
+ Deterministic *shape regex + checksum* — the checksum rejects look-alikes so the
81
+ guard never trips on a random 13-digit string. Separators are stripped first, so
82
+ `756.1234.5678.97` and `756 1234 5678 97` validate identically.
83
+
84
+ | Category | Identifier | Validation |
85
+ |---|---|---|
86
+ | `ch_ahv` | Swiss AHV / AVS number | EAN-13 check digit |
87
+ | `iban` | CH / LI IBAN | ISO-7064 mod-97 |
88
+ | `credit_card` | Card PAN | Luhn |
89
+ | `ch_phone` | Swiss phone | shape only |
90
+ | `email` | Email | shape only |
91
+ | `dob` | Date of birth | off by default (bare dates false-positive) |
92
+
93
+ **Scope: structured identifiers only.** Person names and street addresses are
94
+ *not* detected — they need an NER model, which would forfeit the deterministic,
95
+ zero-dependency guarantee. Plug your own via `SovereignShield(extra_detectors=[...])`
96
+ (see `SpanDetector`); overlapping spans are dropped fail-closed.
97
+
98
+ **Not encoding-robust.** A model that base64s or ciphers an identifier defeats the
99
+ regex. Separator/whitespace reformatting is handled; encoding is not.
100
+
101
+ ## How it works
102
+
103
+ The thesis, proven in the [K.E.V.I.N.](https://github.com/acoseac/kevin) red-team
104
+ research this is extracted from: you can't close a data leak from *inside* the
105
+ model — a jailbreak, a pretext, or a forced output schema will make it disclose.
106
+ So you put a **deterministic, offline boundary** around the model instead. Sovereign
107
+ Shield is that boundary, as a library: detect → tokenize → (model) → restore.
108
+
109
+ The browser demo ships a TypeScript port of the exact same detectors, kept
110
+ **byte-for-byte in parity** with this Python source by a generated vector suite —
111
+ so redaction on the client and on the server can never silently drift.
112
+
113
+ ## Development
114
+
115
+ ```bash
116
+ pip install -e ".[dev]"
117
+ pytest # unit + round-trip suite
118
+ ruff check . && ruff format --check . && mypy
119
+ python scripts/gen_shield_vectors.py --check # Python parity vectors current
120
+ cd web && npm install && npm run parity # TS shield reproduces them exactly
121
+ ```
122
+
123
+ The `web/` directory is the live demo (Next.js). See [web/README.md](web/README.md).
124
+
125
+ ## Credits & license
126
+
127
+ Extracted from the [K.E.V.I.N.](https://github.com/acoseac/kevin) adversarial-testing
128
+ project; background in the [FADP AI-gateway write-up](https://www.ars.md/blog).
129
+ Licensed under [Apache-2.0](LICENSE).
@@ -0,0 +1,54 @@
1
+ # Releasing
2
+
3
+ `sovereign-shield-ch` publishes to PyPI via **Trusted Publishing** (OIDC) — no API
4
+ token or secret is stored in the repo. The [`Release`](.github/workflows/release.yml)
5
+ workflow builds the sdist + wheel, runs `twine check --strict`, and uploads
6
+ whenever a **GitHub Release** is published.
7
+
8
+ ## One-time setup (before the first release)
9
+
10
+ The PyPI project does not exist yet, so register a **pending publisher** at
11
+ <https://pypi.org/manage/account/publishing/> with these exact values:
12
+
13
+ | Field | Value |
14
+ |---|---|
15
+ | PyPI Project Name | `sovereign-shield-ch` |
16
+ | Owner | `acoseac` |
17
+ | Repository name | `sovereign-shield` |
18
+ | Workflow name | `release.yml` |
19
+ | Environment name | `pypi` |
20
+
21
+ Then, in the GitHub repo, create an **environment** named `pypi`
22
+ (Settings → Environments → New environment). Optionally add yourself as a
23
+ required reviewer so a publish can't run without a manual approval — the workflow
24
+ already references `environment: pypi`.
25
+
26
+ (Optional) Register the same as a pending publisher on
27
+ <https://test.pypi.org/manage/account/publishing/> if you want to rehearse
28
+ against TestPyPI first.
29
+
30
+ ## Cutting a release
31
+
32
+ 1. Bump the version — **single source of truth** is `__version__` in
33
+ [`src/sovereign_shield/__init__.py`](src/sovereign_shield/__init__.py)
34
+ (hatchling reads it; `pyproject.toml` has no separate version to sync).
35
+ 2. Commit and merge to `main` (CI green).
36
+ 3. Tag and publish a GitHub Release matching the version:
37
+ ```bash
38
+ gh release create v0.1.0 --title v0.1.0 --generate-notes
39
+ ```
40
+ 4. The `Release` workflow runs automatically: build → `twine check --strict` →
41
+ publish to PyPI via OIDC. Watch it under the **Actions** tab (and approve the
42
+ `pypi` environment if you enabled a required reviewer).
43
+
44
+ ## Local dry run
45
+
46
+ Reproduce exactly what CI does, without publishing:
47
+
48
+ ```bash
49
+ pip install -e ".[dev]" build twine
50
+ python -m build
51
+ twine check --strict dist/*
52
+ # inspect the wheel is package-only (no web/ demo):
53
+ python -m zipfile -l dist/*.whl
54
+ ```