proxy2model 2.0.0__tar.gz → 2.1.0__tar.gz

{proxy2model-2.0.0/src/proxy2model.egg-info → proxy2model-2.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: proxy2model
-Version: 2.0.0
+Version: 2.1.0
 Summary: Turn captured proxy traffic into an evidence-backed model of a web app.
 Author: falc0n
 License-Expression: Apache-2.0
@@ -39,43 +39,89 @@ Dynamic: license-file
 
 ### `//` the instrument
 
-Mapping an app by hand takes hours, and it starts over on every engagement: which hosts, pages and
-endpoints exist, which page calls what, what the JavaScript names but you never reached, what each
-role can see. **proxy2model does that bookkeeping from the traffic you already capture**, so the time
-goes to testing. It does not scan or attack, and it never calls anything a vulnerability. It tells
-you what is there, which request proves it, and what the capture could not show.
+**Every claim cites the request that proves it, and the model names what the capture could not see.**
+proxy2model turns the traffic you already capture into a model of the web app: its endpoints, pages,
+parameters and roles. It compares what each role reached and hands an AI cited evidence instead of
+raw traffic. It does not scan or attack, and it never calls anything a vulnerability: it gives
+hypotheses and the next safe check.
 
-[![license](https://img.shields.io/badge/license-apache--2.0-e85002?style=flat-square&labelColor=0b0a08)](https://github.com/falc0n-researcher/proxy2model/blob/main/LICENSE)
+[![pypi](https://img.shields.io/pypi/v/proxy2model?style=flat-square&labelColor=0b0a08&color=e85002)](https://pypi.org/project/proxy2model/)
 [![tests](https://img.shields.io/github/actions/workflow/status/falc0n-researcher/proxy2model/ci.yml?branch=main&style=flat-square&label=tests&labelColor=0b0a08&color=ded2c0)](https://github.com/falc0n-researcher/proxy2model/actions/workflows/ci.yml)
 [![secrets](https://img.shields.io/badge/secrets-masked%20before%20write-ded2c0?style=flat-square&labelColor=0b0a08)](#-boundaries)
-![runtime](https://img.shields.io/badge/python%203.10%2B-no%20dependencies-6f6a61?style=flat-square&labelColor=0b0a08)
+[![license](https://img.shields.io/badge/license-apache--2.0-6f6a61?style=flat-square&labelColor=0b0a08)](https://github.com/falc0n-researcher/proxy2model/blob/main/LICENSE)
+
+```bash
+git clone https://github.com/falc0n-researcher/proxy2model && cd proxy2model && ./demo.sh
+```
+
+Sixty seconds, nothing to install, no network: four recorded OWASP Juice Shop sessions are modelled,
+anonymous is compared with a signed-in user, the surface the login opened is diffed, and the report opens.
 
 ---
 
-### `//` field test — the bundled capture
+### `//` cross-role — what each role should not have reached
 
-<img src="https://raw.githubusercontent.com/falc0n-researcher/proxy2model/main/assets/readme/ledger.svg" alt="From the Juice Shop sample: 104 endpoints mapped, 16 named in code but never called, 79 secret values masked" width="100%">
+<img src="https://raw.githubusercontent.com/falc0n-researcher/proxy2model/main/assets/readme/ledger.svg" alt="From the Juice Shop sample: an admin endpoint answered anonymous with 200, anonymous changed a product with PUT, and context.json is 37k tokens instead of about 440k of raw traffic" width="100%">
 
-Four [OWASP Juice Shop](https://owasp.org/www-project-juice-shop/) sessions (417 requests as anonymous,
-user, accounting and admin), modelled in about two seconds on a laptop:
+Record the app once per role (`--role anon`, `--role user`, ...) and ask what the lower one reached:
 
-- **001 · The API, mapped** — 104 endpoints, and the observed API as OpenAPI 3.0: 68 paths, 88 operations, ready for Postman, ZAP or a fuzzer.
-- **002 · Named, never called** — 16 endpoints the app's JavaScript names but the browsing never reached: `/rest/admin`, `/rest/chat`, `/rest/track-order`, `/api/SecurityAnswers` ...
-- **003 · Anonymous got 200 from admin-looking endpoints**, cited to the request:
-  ```
-  $ proxy2model cross-role juiceshop --low anon --high user
-  'anon' reached privileged-looking endpoints (2)
-    GET /rest/admin/application-configuration    ev_93  anon=[200] user=[200]
-      unknown: whether the low-role response was authorised or a leak · next: compare 'anon' and 'user' response bodies
-  ```
+```
+$ proxy2model cross-role juiceshop --low anon --high user
+anon vs user: 2 privileged-looking reached · 3 object reads to test · 2 writes by anon · 31 untested · 12 enforced
 
-Every line cites the request behind it (`ev_93`), and every lead says what is still unknown and the
-next safe check. Reproduce it:
+'anon' reached privileged-looking endpoints (2)  [function-level authorization (OWASP API5)]
+  GET /rest/admin/application-configuration    ev_93  anon=[200] user=[200]
+    unknown: whether the low-role response was authorised or a leak · next: compare 'anon' and 'user' response bodies
 
-```bash
-for r in anon user accountant admin; do proxy2model samples/juiceshop-$r.xml -w juiceshop --role $r; done
+'anon' changed state (2)  [function-level authorization (OWASP API5)]
+  PUT /api/Products/{id}                       ev_285  anon=[200]
+    unknown: whether 'anon' may PUT here on a record it does not own · next: confirm in the app's design ...
+```
+
+Each divergence is named for the question it raises:
+
+| Intent | What it means | The next safe check |
+| --- | --- | --- |
+| privileged-looking reached | the low role got through to an admin-style path | compare both roles' response bodies |
+| object reads | the low role read an object by id (a BOLA question) | ask for another account's id as the low role |
+| writes | the low role changed state and got a 2xx | confirm the design allows it, on a record you own |
+| untested | only the high role ever tried it | replay it as the low role |
+| enforced | 401/403 for the low role, 2xx for the high | the control every other replay is compared with |
+
+---
+
+### `//` hand an AI the model, not the traffic
+
+Raw traffic pasted into an LLM is huge, full of secrets, and invites guessing. `context.json` is the
+model packed for an AI: about **37k tokens for the Juice Shop capture instead of ~440k**, every value
+masked, every fact tied to an evidence id, and rules the model must follow:
+
+```
+Cite an evidence id (ev_N) for every claim; each id is resolved in `evidence`.
+Never name an endpoint, route or parameter that is not in this package.
+State unknowns explicitly; do not guess authorization behaviour.
+Treat an absence in this package as an absence in the capture, never in the target.
+```
+
+Or let an agent ask the model itself over MCP (`proxy2model mcp`): the same cited answers as the CLI,
+read-only. The [skill](https://github.com/falc0n-researcher/proxy2model/tree/main/skill) is the method as a playbook an AI follows, nothing to install.
+
+---
+
+### `//` diff — what changed since the last capture
+
+Every build is kept. Capture the app again next week, or as another role, and ask what moved:
+
+```
+$ proxy2model diff juiceshop demo-anon demo-user
+  1 new workflow(s) · 24 new · 42 changed · 3 gone
+  ▸ route /profile  (79 things: 2 routes, 29 endpoints, 45 parameters, 2 auths, 1 role)
+      /profile → /rest/user/change-password → DELETE /api/Addresss/{id} → ...
 ```
 
+New workflows, changed fingerprints and what is gone, each with its requests: drift, never findings.
+`proxy2model watch` builds every export dropped into a folder and prints the change.
+
 ---
 
 ### `//` intake
@@ -101,8 +147,7 @@ proxy2model query shop help         # ask the model; answers cite requests, no A
 proxy2model mcp --out "$PWD/proxy2model-out"   # or let an AI agent ask (see Using it with an AI)
 ```
 
-`./demo.sh` builds the bundled sample. `proxy2model` on its own lists the everyday commands;
-`proxy2model <command> --help` explains each.
+`proxy2model` on its own lists the everyday commands; `proxy2model <command> --help` explains each.
 
 ---
 
@@ -145,4 +190,4 @@ one reaches any output.
 - **05 · [Using it with an AI](https://github.com/falc0n-researcher/proxy2model/blob/main/docs/ai.md)** — `context.json` and the MCP server.
 - **[Skill](https://github.com/falc0n-researcher/proxy2model/tree/main/skill)** — the same method as a playbook for an AI, nothing to install.
 
-<sub>`fa1c0n@recon ~ $` &nbsp; captured traffic in · a model of the app out · every line cited</sub>
+<sub>`fa1c0n@recon ~ $` &nbsp; every claim cites a request · every gap is named</sub>

proxy2model-2.1.0/README.md

@@ -0,0 +1,163 @@
+<!-- ─────────────────────────────────────────────────────────────
+     proxy2model · README
+     The artwork in assets/readme/ is self-contained SVG (Nippo embedded).
+     Do not edit it by hand: change scripts/build_readme_svgs.py and run it.
+     Links are absolute so the same file renders on PyPI.
+     ───────────────────────────────────────────────────────────── -->
+
+<img src="https://raw.githubusercontent.com/falc0n-researcher/proxy2model/main/assets/readme/hero.svg" alt="proxy2model: every endpoint, page and role you captured, each tied to its request" width="100%">
+
+### `//` the instrument
+
+**Every claim cites the request that proves it, and the model names what the capture could not see.**
+proxy2model turns the traffic you already capture into a model of the web app: its endpoints, pages,
+parameters and roles. It compares what each role reached and hands an AI cited evidence instead of
+raw traffic. It does not scan or attack, and it never calls anything a vulnerability: it gives
+hypotheses and the next safe check.
+
+[![pypi](https://img.shields.io/pypi/v/proxy2model?style=flat-square&labelColor=0b0a08&color=e85002)](https://pypi.org/project/proxy2model/)
+[![tests](https://img.shields.io/github/actions/workflow/status/falc0n-researcher/proxy2model/ci.yml?branch=main&style=flat-square&label=tests&labelColor=0b0a08&color=ded2c0)](https://github.com/falc0n-researcher/proxy2model/actions/workflows/ci.yml)
+[![secrets](https://img.shields.io/badge/secrets-masked%20before%20write-ded2c0?style=flat-square&labelColor=0b0a08)](#-boundaries)
+[![license](https://img.shields.io/badge/license-apache--2.0-6f6a61?style=flat-square&labelColor=0b0a08)](https://github.com/falc0n-researcher/proxy2model/blob/main/LICENSE)
+
+```bash
+git clone https://github.com/falc0n-researcher/proxy2model && cd proxy2model && ./demo.sh
+```
+
+Sixty seconds, nothing to install, no network: four recorded OWASP Juice Shop sessions are modelled,
+anonymous is compared with a signed-in user, the surface the login opened is diffed, and the report opens.
+
+---
+
+### `//` cross-role — what each role should not have reached
+
+<img src="https://raw.githubusercontent.com/falc0n-researcher/proxy2model/main/assets/readme/ledger.svg" alt="From the Juice Shop sample: an admin endpoint answered anonymous with 200, anonymous changed a product with PUT, and context.json is 37k tokens instead of about 440k of raw traffic" width="100%">
+
+Record the app once per role (`--role anon`, `--role user`, ...) and ask what the lower one reached:
+
+```
+$ proxy2model cross-role juiceshop --low anon --high user
+anon vs user: 2 privileged-looking reached · 3 object reads to test · 2 writes by anon · 31 untested · 12 enforced
+
+'anon' reached privileged-looking endpoints (2)  [function-level authorization (OWASP API5)]
+  GET /rest/admin/application-configuration    ev_93  anon=[200] user=[200]
+    unknown: whether the low-role response was authorised or a leak · next: compare 'anon' and 'user' response bodies
+
+'anon' changed state (2)  [function-level authorization (OWASP API5)]
+  PUT /api/Products/{id}                       ev_285  anon=[200]
+    unknown: whether 'anon' may PUT here on a record it does not own · next: confirm in the app's design ...
+```
+
+Each divergence is named for the question it raises:
+
+| Intent | What it means | The next safe check |
+| --- | --- | --- |
+| privileged-looking reached | the low role got through to an admin-style path | compare both roles' response bodies |
+| object reads | the low role read an object by id (a BOLA question) | ask for another account's id as the low role |
+| writes | the low role changed state and got a 2xx | confirm the design allows it, on a record you own |
+| untested | only the high role ever tried it | replay it as the low role |
+| enforced | 401/403 for the low role, 2xx for the high | the control every other replay is compared with |
+
+---
+
+### `//` hand an AI the model, not the traffic
+
+Raw traffic pasted into an LLM is huge, full of secrets, and invites guessing. `context.json` is the
+model packed for an AI: about **37k tokens for the Juice Shop capture instead of ~440k**, every value
+masked, every fact tied to an evidence id, and rules the model must follow:
+
+```
+Cite an evidence id (ev_N) for every claim; each id is resolved in `evidence`.
+Never name an endpoint, route or parameter that is not in this package.
+State unknowns explicitly; do not guess authorization behaviour.
+Treat an absence in this package as an absence in the capture, never in the target.
+```
+
+Or let an agent ask the model itself over MCP (`proxy2model mcp`): the same cited answers as the CLI,
+read-only. The [skill](https://github.com/falc0n-researcher/proxy2model/tree/main/skill) is the method as a playbook an AI follows, nothing to install.
+
+---
+
+### `//` diff — what changed since the last capture
+
+Every build is kept. Capture the app again next week, or as another role, and ask what moved:
+
+```
+$ proxy2model diff juiceshop demo-anon demo-user
+  1 new workflow(s) · 24 new · 42 changed · 3 gone
+  ▸ route /profile  (79 things: 2 routes, 29 endpoints, 45 parameters, 2 auths, 1 role)
+      /profile → /rest/user/change-password → DELETE /api/Addresss/{id} → ...
+```
+
+New workflows, changed fingerprints and what is gone, each with its requests: drift, never findings.
+`proxy2model watch` builds every export dropped into a folder and prints the change.
+
+---
+
+### `//` intake
+
+| Traffic from | Command |
+| --- | --- |
+| Burp (*Save items* XML), Logger++ CSV, Caido (JSON or CSV), or a HAR (browser, ZAP, mitmproxy) | `proxy2model FILE -w app` |
+| A browser it opens and records while you use the app | `proxy2model browse` |
+| A crawl of a running app | `proxy2model crawl URL -w app --yes` |
+
+Every value is masked as it is read. One model comes out: a report, a SQLite graph you can query,
+OpenAPI, and a compact package for an AI. Runs locally.
+
+```bash
+pip install proxy2model
+```
+
+```bash
+proxy2model history.xml -w shop     # build from an export
+proxy2model browse                  # or open a browser and use the app (asks for URL, scope, mode)
+
+proxy2model query shop help         # ask the model; answers cite requests, no AI involved
+proxy2model mcp --out "$PWD/proxy2model-out"   # or let an AI agent ask (see Using it with an AI)
+```
+
+`proxy2model` on its own lists the everyday commands; `proxy2model <command> --help` explains each.
+
+---
+
+### `//` what comes out
+
+```
+proxy2model-out/shop/
+├── report.html     interactive report: overview, what to check, endpoints, map, every request, query console
+├── model.json      the model, and the request behind every claim
+├── openapi.json    the observed API, OpenAPI 3.0
+├── context.json    what to hand an AI instead of raw traffic
+├── graph.db        SQLite graph, queried with `proxy2model q`
+├── exports/        graph.json · graph.graphml · build.svg
+└── inputs/         masked records of every capture, merged on rebuild
+```
+
+---
+
+### `//` boundaries
+
+| It does | It never |
+| --- | --- |
+| Read traffic you captured | Upload, forward or replay it |
+| Mask every value before the first write | Store a token, cookie, key or PII value |
+| Say what it could not observe | Claim coverage it did not have |
+| Offer hypotheses and the evidence to test them | Call anything a vulnerability |
+
+A plain build is offline. `crawl`, `browse --active` and `osint` send requests of their own: use
+them only on systems you are authorized to test. CI plants fake secrets in the samples and fails if
+one reaches any output.
+
+---
+
+### `//` field manual
+
+- **01 · [Capturing traffic](https://github.com/falc0n-researcher/proxy2model/blob/main/docs/capture.md)** — Burp, Caido, HAR, `browse`, roles.
+- **02 · [Crawling a running app](https://github.com/falc0n-researcher/proxy2model/blob/main/docs/crawling.md)** — `crawl` and scripted journeys.
+- **03 · [How it works](https://github.com/falc0n-researcher/proxy2model/blob/main/docs/model.md)** — what is in the model and how it is built.
+- **04 · [Querying the model](https://github.com/falc0n-researcher/proxy2model/blob/main/docs/querying.md)** — `query`, BQL, graph reach, diffs between builds.
+- **05 · [Using it with an AI](https://github.com/falc0n-researcher/proxy2model/blob/main/docs/ai.md)** — `context.json` and the MCP server.
+- **[Skill](https://github.com/falc0n-researcher/proxy2model/tree/main/skill)** — the same method as a playbook for an AI, nothing to install.
+
+<sub>`fa1c0n@recon ~ $` &nbsp; every claim cites a request · every gap is named</sub>

{proxy2model-2.0.0 → proxy2model-2.1.0}/assets/readme/hero.svg

@@ -23,8 +23,8 @@
 <rect x="72" y="345" width="54" height="20" rx="4" fill="none" stroke="#e85002" stroke-opacity="0.5"/>
 <circle cx="83" cy="355" r="3" fill="#e85002"><animate attributeName="opacity" values="1;0.25;1" dur="1.8s" repeatCount="indefinite"/></circle>
 <text x="91" y="359" fill="#e85002" font-size="10" font-weight="500" letter-spacing="1.5px">LOCAL</text>
-<text x="142" y="361" font-size="14.5"><tspan fill="#e85002">fa1c0n@recon ~ $</tspan><tspan fill="#ded2c0"> proxy2model juiceshop-*.xml -w juiceshop</tspan></text>
-<text x="1128" y="361" text-anchor="end" font-size="14.5" font-weight="500"><tspan fill="#f6f1e9">104 endpoints</tspan><tspan fill="#6f6a61"> &#183; </tspan><tspan fill="#f6f1e9">16 never called</tspan></text>
+<text x="142" y="361" font-size="14.5"><tspan fill="#e85002">fa1c0n@recon ~ $</tspan><tspan fill="#ded2c0"> proxy2model cross-role juiceshop --low anon --high user</tspan></text>
+<text x="1128" y="361" text-anchor="end" font-size="14.5" font-weight="500"><tspan fill="#f6f1e9">2 admin paths</tspan><tspan fill="#6f6a61"> &#183; </tspan><tspan fill="#f6f1e9">2 anon writes</tspan></text>
 <g transform="translate(1000,22) scale(1.9)"><path d="M37 39.5 L72 39.5 L26.5 66.5 Q43 51 46 49 Q39 45 37 39.5 Z" fill="#e85002"/></g>
 </g>
 </svg>

{proxy2model-2.0.0 → proxy2model-2.1.0}/assets/readme/ledger.svg

@@ -11,7 +11,7 @@
   <clipPath id="round"><rect x="1" y="1" width="1198" height="248" rx="14"/></clipPath>
 </defs>
 <rect width="1200" height="250" fill="none"/>
-<g><rect x="0" y="2" width="385.3333333333333" height="234" rx="9" fill="#100e0b" stroke="#ded2c0" stroke-opacity="0.12"/><rect x="0" y="2" width="385.3333333333333" height="234" rx="9" fill="url(#sheen)"/><text x="26" y="40" fill="#e85002" font-size="12" font-weight="500" letter-spacing="1.5px">001</text><text x="60" y="40" fill="#6f6a61" font-size="10.5" letter-spacing="2px">THE&#160;API,&#160;MAPPED</text><circle cx="359.3333333333333" cy="36" r="3.5" fill="#6f6a61"/><text x="26" y="78" fill="#f6f1e9" font-size="22" font-weight="500" letter-spacing="-0.3px">Juice Shop, four roles</text><text x="26" y="150" fill="#f3ece0" font-size="52" font-weight="200" letter-spacing="-1px">104</text><text x="26" y="180" fill="#a39a8d" font-size="13" font-weight="400">endpoints from 417 captured</text><text x="26" y="199" fill="#a39a8d" font-size="13" font-weight="400">requests, each with its proof</text><line x1="26" y1="200" x2="359.3333333333333" y2="200" stroke="#ded2c0" stroke-opacity="0.08"/><rect x="26.0" y="206" width="33.6" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="42.8" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">API</text><rect x="66.6" y="206" width="62.4" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="97.8" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">OPENAPI</text><rect x="136.0" y="206" width="48.0" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="160.0" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">ROLES</text><text x="359.3333333333333" y="220" text-anchor="end" fill="#6f6a61" font-size="10" letter-spacing="1.5px">68&#160;PATHS&#160;·&#160;88&#160;OPS</text></g>
-<g><rect x="407.3333333333333" y="2" width="385.3333333333333" height="234" rx="9" fill="#100e0b" stroke="#e85002" stroke-opacity="0.55"/><rect x="407.3333333333333" y="2" width="385.3333333333333" height="234" rx="9" fill="url(#sheen)"/><rect x="407.3333333333333" y="2" width="385.3333333333333" height="3" rx="1.5" fill="#e85002"/><text x="433.3333333333333" y="40" fill="#e85002" font-size="12" font-weight="500" letter-spacing="1.5px">002</text><text x="467.3333333333333" y="40" fill="#6f6a61" font-size="10.5" letter-spacing="2px">WHAT&#160;NOBODY&#160;WALKED</text><circle cx="766.6666666666666" cy="36" r="3.5" fill="#e85002"/><text x="433.3333333333333" y="78" fill="#f6f1e9" font-size="22" font-weight="500" letter-spacing="-0.3px">Named, never called</text><text x="433.3333333333333" y="150" fill="#e85002" font-size="52" font-weight="200" letter-spacing="-1px">16</text><text x="433.3333333333333" y="180" fill="#a39a8d" font-size="13" font-weight="400">endpoints the JavaScript names</text><text x="433.3333333333333" y="199" fill="#a39a8d" font-size="13" font-weight="400">that the browsing never reached</text><line x1="433.3333333333333" y1="200" x2="766.6666666666666" y2="200" stroke="#ded2c0" stroke-opacity="0.08"/><rect x="433.3" y="206" width="40.8" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="453.7" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">CODE</text><rect x="481.1" y="206" width="48.0" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="505.1" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">LEADS</text><text x="766.6666666666666" y="220" text-anchor="end" fill="#6f6a61" font-size="10" letter-spacing="1.5px">E.G.&#160;/REST/ADMIN</text></g>
-<g><rect x="814.6666666666666" y="2" width="385.3333333333333" height="234" rx="9" fill="#100e0b" stroke="#ded2c0" stroke-opacity="0.12"/><rect x="814.6666666666666" y="2" width="385.3333333333333" height="234" rx="9" fill="url(#sheen)"/><text x="840.6666666666666" y="40" fill="#e85002" font-size="12" font-weight="500" letter-spacing="1.5px">003</text><text x="874.6666666666666" y="40" fill="#6f6a61" font-size="10.5" letter-spacing="2px">MASKED&#160;AT&#160;READ</text><circle cx="1174.0" cy="36" r="3.5" fill="#6f6a61"/><text x="840.6666666666666" y="78" fill="#f6f1e9" font-size="22" font-weight="500" letter-spacing="-0.3px">Proof without secrets</text><text x="840.6666666666666" y="150" fill="#f3ece0" font-size="52" font-weight="200" letter-spacing="-1px">79</text><text x="840.6666666666666" y="180" fill="#a39a8d" font-size="13" font-weight="400">token, cookie and PII values</text><text x="840.6666666666666" y="199" fill="#a39a8d" font-size="13" font-weight="400">masked before the first write</text><line x1="840.6666666666666" y1="200" x2="1174.0" y2="200" stroke="#ded2c0" stroke-opacity="0.08"/><rect x="840.7" y="206" width="55.2" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="868.3" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">TOKENS</text><rect x="902.9" y="206" width="33.6" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="919.7" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">PII</text><text x="1174.0" y="220" text-anchor="end" fill="#6f6a61" font-size="10" letter-spacing="1.5px">0&#160;WRITTEN</text></g>
+<g><rect x="0" y="2" width="385.3333333333333" height="234" rx="9" fill="#100e0b" stroke="#ded2c0" stroke-opacity="0.12"/><rect x="0" y="2" width="385.3333333333333" height="234" rx="9" fill="url(#sheen)"/><text x="26" y="40" fill="#e85002" font-size="12" font-weight="500" letter-spacing="1.5px">001</text><text x="60" y="40" fill="#6f6a61" font-size="10.5" letter-spacing="2px">CROSS-ROLE</text><circle cx="359.3333333333333" cy="36" r="3.5" fill="#6f6a61"/><text x="26" y="78" fill="#f6f1e9" font-size="22" font-weight="500" letter-spacing="-0.3px">Admin answered anonymous</text><text x="26" y="150" fill="#f3ece0" font-size="52" font-weight="200" letter-spacing="-1px">200</text><text x="26" y="180" fill="#a39a8d" font-size="13" font-weight="400">from /rest/admin/application-</text><text x="26" y="199" fill="#a39a8d" font-size="13" font-weight="400">configuration, request cited</text><line x1="26" y1="200" x2="359.3333333333333" y2="200" stroke="#ded2c0" stroke-opacity="0.08"/><rect x="26.0" y="206" width="48.0" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="50.0" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">ROLES</text><rect x="81.0" y="206" width="40.8" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="101.4" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">API5</text><text x="359.3333333333333" y="220" text-anchor="end" fill="#6f6a61" font-size="10" letter-spacing="1.5px">EV_93</text></g>
+<g><rect x="407.3333333333333" y="2" width="385.3333333333333" height="234" rx="9" fill="#100e0b" stroke="#e85002" stroke-opacity="0.55"/><rect x="407.3333333333333" y="2" width="385.3333333333333" height="234" rx="9" fill="url(#sheen)"/><rect x="407.3333333333333" y="2" width="385.3333333333333" height="3" rx="1.5" fill="#e85002"/><text x="433.3333333333333" y="40" fill="#e85002" font-size="12" font-weight="500" letter-spacing="1.5px">002</text><text x="467.3333333333333" y="40" fill="#6f6a61" font-size="10.5" letter-spacing="2px">CROSS-ROLE</text><circle cx="766.6666666666666" cy="36" r="3.5" fill="#e85002"/><text x="433.3333333333333" y="78" fill="#f6f1e9" font-size="22" font-weight="500" letter-spacing="-0.3px">Anonymous changed a product</text><text x="433.3333333333333" y="150" fill="#e85002" font-size="52" font-weight="200" letter-spacing="-1px">PUT</text><text x="433.3333333333333" y="180" fill="#a39a8d" font-size="13" font-weight="400">/api/Products/{id} returned 200</text><text x="433.3333333333333" y="199" fill="#a39a8d" font-size="13" font-weight="400">with no login at all</text><line x1="433.3333333333333" y1="200" x2="766.6666666666666" y2="200" stroke="#ded2c0" stroke-opacity="0.08"/><rect x="433.3" y="206" width="48.0" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="457.3" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">ROLES</text><rect x="488.3" y="206" width="40.8" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="508.7" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">API5</text><text x="766.6666666666666" y="220" text-anchor="end" fill="#6f6a61" font-size="10" letter-spacing="1.5px">EV_285</text></g>
+<g><rect x="814.6666666666666" y="2" width="385.3333333333333" height="234" rx="9" fill="#100e0b" stroke="#ded2c0" stroke-opacity="0.12"/><rect x="814.6666666666666" y="2" width="385.3333333333333" height="234" rx="9" fill="url(#sheen)"/><text x="840.6666666666666" y="40" fill="#e85002" font-size="12" font-weight="500" letter-spacing="1.5px">003</text><text x="874.6666666666666" y="40" fill="#6f6a61" font-size="10.5" letter-spacing="2px">CONTEXT.JSON</text><circle cx="1174.0" cy="36" r="3.5" fill="#6f6a61"/><text x="840.6666666666666" y="78" fill="#f6f1e9" font-size="22" font-weight="500" letter-spacing="-0.3px">Hand an AI the model</text><text x="840.6666666666666" y="150" fill="#f3ece0" font-size="52" font-weight="200" letter-spacing="-1px">37k</text><text x="840.6666666666666" y="180" fill="#a39a8d" font-size="13" font-weight="400">tokens of cited evidence instead</text><text x="840.6666666666666" y="199" fill="#a39a8d" font-size="13" font-weight="400">of ~440k of raw traffic</text><line x1="840.6666666666666" y1="200" x2="1174.0" y2="200" stroke="#ded2c0" stroke-opacity="0.08"/><rect x="840.7" y="206" width="26.4" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="853.9" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">AI</text><rect x="874.1" y="206" width="33.6" height="18" rx="3" fill="none" stroke="#ded2c0" stroke-opacity="0.16"/><text x="890.9" y="219" fill="#a39a8d" font-size="9.5" font-weight="400" text-anchor="middle" letter-spacing="1px">MCP</text><text x="1174.0" y="220" text-anchor="end" fill="#6f6a61" font-size="10" letter-spacing="1.5px">SECRETS&#160;MASKED</text></g>
 </svg>

{proxy2model-2.0.0 → proxy2model-2.1.0}/src/proxy2model/__init__.py

@@ -1,5 +1,5 @@
 """proxy2model — turn captured proxy traffic into an evidence-backed web-app model."""
-__version__ = "2.0.0"
+__version__ = "2.1.0"
 
 import os as _os
 

{proxy2model-2.0.0 → proxy2model-2.1.0}/src/proxy2model/cli/analyze.py

@@ -96,19 +96,31 @@ def cmd_crossrole(args) -> int:
         print(json.dumps(result, indent=2))
         return 0
 
-    def show(title, items):
-        print(f"{title} ({len(items)})")
-        for s in items:
+    lo, hi, h = args.low, args.high, result["headline"]
+    print(f"{lo} vs {hi}: {h['low_reached_privileged']} privileged-looking reached · "
+          f"{h['object_reads']} object reads to test · {h['low_writes']} writes by {lo} · "
+          f"{h['high_only']} untested · {h['enforced']} enforced\n")
+
+    def show(title, items, limit=None):
+        if not items:
+            return
+        print(f"{title} ({len(items)})  [{items[0]['family']}]")
+        for s in items[:limit]:
             ev = ", ".join(f"ev_{e}" for e in s["evidence"])
             st = " ".join(f"{r}={v}" for r, v in s["statuses"].items())
             print(f"  {s['endpoint']:<44} {ev}  {st}")
             print(f"    unknown: {s['unknown']} · next: {s['next']}")
+        if limit and len(items) > limit:
+            print(f"  ... {len(items) - limit} more (--json for all)")
         print()
 
-    show(f"'{args.low}' reached privileged-looking endpoints", result["low_reached_privileged"])
-    show(f"'{args.high}' reached, '{args.low}' never tried", result["high_only"])
-    show("both roles got 2xx", result["shared_same_success"])
-    print("Each is a hypothesis, not a finding.")
+    show(f"'{lo}' reached privileged-looking endpoints", result["low_reached_privileged"])
+    show(f"'{lo}' read objects by id", result["object_reads"])
+    show(f"'{lo}' changed state", result["low_writes"])
+    show(f"'{hi}' reached, '{lo}' never tried", result["high_only"], limit=None if args.all else 10)
+    show(f"denied to '{lo}', allowed to '{hi}'", result["enforced"])
+    show("both roles got 2xx", result["shared_same_success"], limit=None if args.all else 10)
+    print("Each is a hypothesis to verify in an authorised session, not a finding.")
     return 0
 
 

{proxy2model-2.0.0 → proxy2model-2.1.0}/src/proxy2model/cli/app.py

@@ -214,6 +214,7 @@ def main(argv=None) -> int:
     cr.add_argument("--high", required=True, type=_name("--high"),
                     help="the more-privileged role (as tagged with --role)")
     cr.add_argument("--json", action="store_true", help="print JSON instead of text")
+    cr.add_argument("--all", action="store_true", help="list every entry (long sections stop at 10)")
     cr.add_argument("--out", default="proxy2model-out",
                     help="output directory used at build time")
     cr.set_defaults(func=cmd_crossrole)

{proxy2model-2.0.0 → proxy2model-2.1.0}/src/proxy2model/cli/browse.py

@@ -226,6 +226,7 @@ def cmd_browse(args) -> int:
     if recorder.error:
         print(ui.warn(f"recording stopped early: {recorder.error}"), file=sys.stderr)
     live.flush()
+    live.finish()
     if not live.saved:
         print(f"\n  nothing in scope was recorded ({', '.join(scope)})")
         return 1

{proxy2model-2.0.0 → proxy2model-2.1.0}/src/proxy2model/cli/mcp.py

@@ -32,7 +32,9 @@ TOOLS = [
      "inputSchema": {"type": "object", "required": ["app", "expression"],
                      "properties": {"app": _APP, "expression": {"type": "string"},
                                     "limit": {"type": "integer", "minimum": 1, "maximum": 1000}}}},
-    {"name": "cross_role", "description": "What the lower-privileged role reached that only the higher one should.",
+    {"name": "cross_role", "description": "Authorization hypotheses between two roles: what the lower one reached "
+     "that looks privileged, objects it read by id, state it changed, what it never tried, and where it was denied; "
+     "each with the request that shows it and a safe next check.",
      "inputSchema": {"type": "object", "required": ["app", "low", "high"],
                      "properties": {"app": _APP, "low": {"type": "string"}, "high": {"type": "string"}}}},
     {"name": "builds", "description": "The builds of a model kept in graph.db, to pick two for diff.",

{proxy2model-2.0.0 → proxy2model-2.1.0}/src/proxy2model/core/model.py

@@ -859,16 +859,35 @@ def _ok(statuses) -> bool:
     return any(200 <= s < 300 for s in statuses or [])
 
 
-def cross_role(m: Model, low: str, high: str) -> dict:
-    """Compare what a low and a high role reached. Output is hypotheses.
+# the authorization question each kind of divergence raises, named the way an API tester names it
+CROSS_ROLE_FAMILY = {
+    "low_reached_privileged": "function-level authorization (OWASP API5)",
+    "object_reads": "object-level authorization (OWASP API1)",
+    "low_writes": "function-level authorization (OWASP API5)",
+    "high_only": "untested authorization",
+    "enforced": "a control: authorization held here",
+    "shared_same_success": "data exposure between roles (OWASP API3)",
+}
+_OBJECT_KEY = re.compile(r"\{(?:id|uuid|hash|code|token)\}")
+_WRITE_METHODS = {"POST", "PUT", "PATCH", "DELETE"}
+
 
-    high_only                endpoints the high role reached that the low role
-                             never requested — replay each as the low role.
-    low_reached_privileged   endpoints the low role reached whose path looks
-                             privileged — check the response was meant for it.
-    shared_same_success      both roles got a 2xx — compare response content.
+def cross_role(m: Model, low: str, high: str) -> dict:
+    """Compare what a low and a high role reached. Output is hypotheses, never findings.
+
+    low_reached_privileged   the low role got through to a privileged-looking path.
+    object_reads             the low role got a 2xx from an endpoint keyed by an object id:
+                             ask for the high role's object as the low role (BOLA).
+    low_writes               the low role changed state (POST, PUT, PATCH, DELETE) and got a 2xx.
+    high_only                the high role reached it, the low role never tried: replay it.
+    enforced                 the low role got 401/403 where the high role got 2xx: the
+                             control to compare every other replay against.
+    shared_same_success      both roles got a 2xx: compare what each was shown.
+
+    Each entry carries `family` (the authorization question), `unknown` and a safe `next` check.
     """
     high_only, low_priv, shared = [], [], []
+    objects, writes, enforced = [], [], []
     ev_role = {e["id"]: e.get("role") for e in m.evidence_log.meta()}
 
     def evs(node, prefer_role):
@@ -886,27 +905,51 @@ def cross_role(m: Model, low: str, high: str) -> dict:
         entry = {"endpoint": node.label, "id": node.id,
                  "evidence": evs(node, low if rl else high),
                  "statuses": {r: sbr.get(r, []) for r in (low, high) if r in sbr}}
+        low_ok, high_ok = _ok(sbr.get(low)), _ok(sbr.get(high))
+        method = (node.attrs.get("method") or "").upper()
+        tag = lambda kind, **kw: {**entry, "kind": kind, "family": CROSS_ROLE_FAMILY[kind], **kw}
         if rh and not rl:
-            high_only.append({**entry, "kind": "high_only",
-                              "unknown": f"whether '{low}' can reach it",
-                              "next": f"replay as '{low}' and compare with '{high}'"})
+            high_only.append(tag("high_only",
+                                 unknown=f"whether '{low}' can reach it",
+                                 next=f"replay as '{low}' and compare with '{high}'"))
         elif rl and PRIV_RE.search(path):
-            low_priv.append({**entry, "kind": "low_reached_privileged",
-                             "unknown": "whether the low-role response was authorised or a leak",
-                             "next": f"compare '{low}' and '{high}' response bodies"})
-        elif rl and rh and _ok(sbr.get(low)) and _ok(sbr.get(high)):
-            shared.append({**entry, "kind": "shared_same_success",
-                           "unknown": "whether both roles see the same data",
-                           "next": "diff the two responses for role-specific fields"})
+            low_priv.append(tag("low_reached_privileged",
+                                unknown="whether the low-role response was authorised or a leak",
+                                next=f"compare '{low}' and '{high}' response bodies"))
+        elif rl and rh and low_ok and high_ok:
+            shared.append(tag("shared_same_success",
+                              unknown="whether both roles see the same data",
+                              next="diff the two responses for role-specific fields"))
+        if rl and low_ok and method in _WRITE_METHODS:
+            on_object = bool(_OBJECT_KEY.search(path))
+            writes.append(tag("low_writes",
+                              unknown=f"whether '{low}' may {method} here" + (" on a record it does not own" if on_object else ""),
+                              next=f"confirm in the app's design that '{low}' may {method} here; if not, "
+                                   f"repeat it as '{low}' on a test record you own"))
+        elif rl and low_ok and _OBJECT_KEY.search(path):
+            objects.append(tag("object_reads",
+                               unknown=f"whether '{low}' can open an object that belongs to '{high}'",
+                               next=(f"as '{low}', request an id seen in '{high}' traffic" if rh else
+                                     f"as '{low}', request an id that belongs to another test account")
+                                    + "; another account's data coming back is the gap, a 403/404 is the control working"))
+        if rl and rh and high_ok and any(s in (401, 403) for s in sbr.get(low, [])):
+            enforced.append(tag("enforced",
+                                unknown="whether every route to this data is guarded the same way",
+                                next=f"use this '{low}' response as the denial to compare other replays with"))
     key = lambda e: e["endpoint"]
-    high_only.sort(key=key)
-    low_priv.sort(key=key)
-    shared.sort(key=key)
+    for xs in (high_only, low_priv, shared, objects, writes, enforced):
+        xs.sort(key=key)
     return {"low": low, "high": high,
             "roles_present": sorted(m.roles),
             "surface": low_priv + high_only,
+            "headline": {"low_reached_privileged": len(low_priv), "object_reads": len(objects),
+                         "low_writes": len(writes), "high_only": len(high_only),
+                         "enforced": len(enforced), "shared_same_success": len(shared)},
             "low_reached_privileged": low_priv,
+            "object_reads": objects,
+            "low_writes": writes,
             "high_only": high_only,
+            "enforced": enforced,
             "shared_same_success": shared}