ARSH871-bot · Sep 12, 2026 · Sep 10, 2026 · Sep 10, 2026 · Sep 12, 2026
@@ -0,0 +1,40 @@
+{
+ "US-20": 313,
+ "US-21": 314,
+ "US-22": 315,
+ "US-23": 316,
+ "US-24": 317,
+ "US-25": 318,
+ "US-26": 319,
+ "US-27": 320,
+ "US-28": 78,
+ "US-29": 321,
+ "US-30": 322,
+ "US-31": 323,
+ "US-32": 324,
+ "US-33": 325,
+ "US-34": 326,
+ "US-35": 327,
+ "US-36": 328,
+ "US-37": 329,
+ "US-38": 330,
+ "US-39": 331,
+ "US-40": 332,
+ "US-41": 333,
+ "US-42": 334,
+ "US-43": 335,
+ "US-44": 336,
+ "US-45": 337,
+ "US-46": 338,
+ "US-47": 339,
+ "US-48": 340,
+ "US-49": 341,
+ "US-50": 342,
+ "US-51": 343,
+ "US-52": 344,
+ "US-53": 220,
+ "US-54": 345,
+ "US-55": 346,
+ "US-56": 347,
+ "US-57": 308
+}
@@ -0,0 +1,213 @@
+<!-- GENERATED by tools/make_phase3_backlog.py. Do not hand-edit.
+     The stories live in that script once, and both this document
+     and the GitHub issues are produced from them, so the two
+     cannot disagree. Hand-editing removes that guarantee. -->
+
+# Phase 3 — market readiness
+
+*Everything between Netwise today and a product that could be sold.*
+
+Phase 0 was consolidation. Phase 1 was the four capabilities. Phase 2 was direction only. This is the first backlog written against a different question: not *what would make this a better capstone* but **what would stop a stranger who paid for it from asking for a refund**.
+
+---
+
+## Where these came from
+
+| Range | Source |
+|---|---|
+| `US-20` – `US-39` | **Discussion #296**, unchanged. Twenty stories designed and argued in early September and never turned into work. Their acceptance criteria are copied rather than rewritten. |
+| `US-40` – `US-56` | **New.** Each one closes a gap a paying customer hits in the first week, or one a buyer's security review asks about and we currently cannot answer. |
+
+Three stories already existed as issues and are folded in rather than created again — creating them twice would be this project's own recurring defect, committed by the backlog written to avoid it:
+
+| Story | Existing issue |
+|---|---|
+| `US-28` | [#78](https://github.com/ARSH871-bot/Netwise/issues/78) — Convert my whole pfSense export, including NAT |
+| `US-53` | [#220](https://github.com/ARSH871-bot/Netwise/issues/220) — One real compliance benchmark pack (CIS Cisco IOS) |
+| `US-57` | [#308](https://github.com/ARSH871-bot/Netwise/issues/308) — Coverage on every surface that shows findings, or none |
+
+Nothing here is a feature because it sounds impressive. Every story names the thing that is currently untrue, unmeasured or unreachable.
+
+---
+
+## The distribution, stated rather than arranged
+
+**38 stories. Arsh has 28. Ankeet has 4, Shubham has 3, Samika has 3.**
+
+That is deliberate, and it is a decision rather than an accident, so it is written down where the people affected by it can read it.
+
+**Why.** The remaining calendar is short and most of this milestone is sequential: US-22 waits on US-33, US-45 waits on US-35, US-46 waits on US-22 and US-44. A dependency chain moves at the speed of whoever is slowest to pick up their link. Concentrating the chain on one person removes the hand-off latency, which is the largest single cost in a four-person team working part-time around other papers.
+
+**What each of the other three gets** is a short, self-contained slice inside the vertical they already own, with no story in it that blocks anybody else. They can finish, stop, and not be the reason something else is waiting.
+
+**What it costs, which is the part worth saying out loud.** #249 measured that Arsh authored 61% of pull requests and consumed 62% of the team's review capacity, and — more importantly — that **review, not authorship, is where this project's defects have actually been caught**. The fixture corruption in #304, the student IDs in #311, the stale amendment table, the false claim in the portfolio: every one was found by somebody reading somebody else's work.
+
+So this split does not reduce the team's work. It moves it from authoring to reviewing. **Every story here still needs a current approving review from someone who did not write it (M-1), and new work still voids an approval (M-2 rule 1).** If the other three treat a light authoring load as a light total load, this milestone will ship faster and be worse, and the failure will not be visible until something built on an unreviewed foundation has to be unpicked.
+
+The honest summary: **three people are being asked to write less and read more.** That is the trade, and it is only a good trade if the reading actually happens.
+
+---
+
+## The backlog
+
+### Honesty
+
+*The product's central claim is that it never says “clean” when it means “unchecked”. These are the places that claim is still only half-kept.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-23` | Make `access_control` read my policy | [#316](https://github.com/ARSH871-bot/Netwise/issues/316) | M | Arsh |
+| `US-24` | Show me what Netwise could not read | [#317](https://github.com/ARSH871-bot/Netwise/issues/317) | M | Arsh |
+| `US-26` | Make `routing` read my policy | [#319](https://github.com/ARSH871-bot/Netwise/issues/319) | M | Arsh |
+| `US-40` | Prove that nothing left my machine | [#332](https://github.com/ARSH871-bot/Netwise/issues/332) | M | Arsh |
+| `US-57` | Coverage on every surface that shows findings, or none *(already open)* | [#308](https://github.com/ARSH871-bot/Netwise/issues/308) | M | Arsh |
+
+### Distribution
+
+*Nobody can buy something they cannot install.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-41` | Install Netwise with one command | [#333](https://github.com/ARSH871-bot/Netwise/issues/333) | M | Arsh |
+| `US-42` | Upgrade without losing my data | [#334](https://github.com/ARSH871-bot/Netwise/issues/334) | M | Arsh |
+| `US-49` | Install it with no internet at all | [#341](https://github.com/ARSH871-bot/Netwise/issues/341) | M | Arsh |
+
+### Scale
+
+*Speed was measured in #218 and is not the problem: fifty devices analyse in 6.9 seconds. What `docs/scale.md` found instead is that the analysis does not GROW with the network -- fifty devices in, one device examined. These stories are about that.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-21` | Analyse my whole network at once, not device by device | [#314](https://github.com/ARSH871-bot/Netwise/issues/314) | M | Arsh |
+| `US-43` | Tell me how big a network you can actually handle | [#335](https://github.com/ARSH871-bot/Netwise/issues/335) | L | Arsh |
+| `US-44` | Run a long scan in the background | [#336](https://github.com/ARSH871-bot/Netwise/issues/336) | L | Arsh |
+
+### Multi-tenant
+
+*One person on one laptop is a tool. Several people in one company is a product.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-35` | Sign in, so my scans are mine | [#327](https://github.com/ARSH871-bot/Netwise/issues/327) | L | Samika |
+| `US-45` | Give my team shared access, with roles | [#337](https://github.com/ARSH871-bot/Netwise/issues/337) | L | Arsh |
+| `US-52` | Tell me what you keep, and delete it when I say | [#344](https://github.com/ARSH871-bot/Netwise/issues/344) | M | Arsh |
+
+### Automation
+
+*A scan somebody has to remember to run is a scan that stops happening in week three.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-20` | Check my configs in CI, before the change merges | [#313](https://github.com/ARSH871-bot/Netwise/issues/313) | M | Arsh |
+| `US-46` | Scan on a schedule, without me starting it | [#338](https://github.com/ARSH871-bot/Netwise/issues/338) | M | Arsh |
+| `US-47` | Read my configs from where they already are | [#339](https://github.com/ARSH871-bot/Netwise/issues/339) | M | Arsh |
+| `US-48` | Send findings to the tools my team already uses | [#340](https://github.com/ARSH871-bot/Netwise/issues/340) | M | Arsh |
+
+### Content
+
+*Our rules are ours. A buyer wants a standard's rules, and a way into writing their own.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-34` | Draft a policy from a config I already have | [#326](https://github.com/ARSH871-bot/Netwise/issues/326) | M | Arsh |
+| `US-53` | One real compliance benchmark pack (CIS Cisco IOS) *(already open)* | [#220](https://github.com/ARSH871-bot/Netwise/issues/220) | L | Arsh |
+
+### Remediation
+
+*Finding the problem is half the value. The other half is the line of config that fixes it.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-54` | Give me the fix, not just the problem | [#345](https://github.com/ARSH871-bot/Netwise/issues/345) | L | Arsh |
+
+### Vendors
+
+*Four vendor fixtures exist and three of them -- Arista, Juniper and NX-OS -- already produce a real finding end to end. What is left is the fourth, and saying so anywhere a user can see it.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-27` | Analyse a vendor that is not Cisco, end to end | [#320](https://github.com/ARSH871-bot/Netwise/issues/320) | L | Ankeet |
+| `US-28` | Convert my whole pfSense export, including NAT *(already open)* | [#78](https://github.com/ARSH871-bot/Netwise/issues/78) | L | Ankeet |
+
+### AI
+
+*The layer with the most attention on it and the least measurement behind it.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-25` | Let me ask a follow-up | [#318](https://github.com/ARSH871-bot/Netwise/issues/318) | L | Ankeet |
+| `US-55` | Show me the explanation layer is any good | [#346](https://github.com/ARSH871-bot/Netwise/issues/346) | M | Ankeet |
+
+### Policy
+
+*The user's own rules are a first-class artefact and deserve the same scrutiny we give their configs.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-30` | Tell me if my own policy contradicts itself | [#322](https://github.com/ARSH871-bot/Netwise/issues/322) | M | Shubham |
+| `US-31` | Check a proposed change against my whole policy | [#323](https://github.com/ARSH871-bot/Netwise/issues/323) | M | Shubham |
+| `US-32` | Tell me which of my policy rules never fire | [#324](https://github.com/ARSH871-bot/Netwise/issues/324) | S | Shubham |
+| `US-33` | Recognise the same problem across two scans | [#325](https://github.com/ARSH871-bot/Netwise/issues/325) | M | Arsh |
+
+### UX
+
+*Most of these are cheap. Together they are the difference between a demo and a thing someone uses on a Tuesday.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-22` | Tell me only what changed since the last scan | [#315](https://github.com/ARSH871-bot/Netwise/issues/315) | M | Arsh |
+| `US-36` | Rank by what it would cost me, not just what kind of bug it is | [#328](https://github.com/ARSH871-bot/Netwise/issues/328) | M | Arsh |
+| `US-37` | Give me something I can hand to an auditor | [#329](https://github.com/ARSH871-bot/Netwise/issues/329) | M | Arsh |
+| `US-38` | Let me use it with a keyboard and a screen reader | [#330](https://github.com/ARSH871-bot/Netwise/issues/330) | M | Samika |
+| `US-39` | Show me progress during a long scan | [#331](https://github.com/ARSH871-bot/Netwise/issues/331) | S | Samika |
+| `US-56` | Make my first five minutes work | [#347](https://github.com/ARSH871-bot/Netwise/issues/347) | S | Arsh |
+
+### Trust
+
+*The questions a buyer's security and procurement teams ask before they ask anything about features.*
+
+| | Story | Issue | Size | Owner |
+|---|---|---|---|---|
+| `US-29` | Explain related findings as one story | [#321](https://github.com/ARSH871-bot/Netwise/issues/321) | M | Arsh |
+| `US-50` | Give me an SBOM and a stated dependency policy | [#342](https://github.com/ARSH871-bot/Netwise/issues/342) | S | Arsh |
+| `US-51` | Point Netwise at its own deployment | [#343](https://github.com/ARSH871-bot/Netwise/issues/343) | S | Arsh |
+
+---
+
+## Order of work
+
+Not a schedule. A dependency order, so nothing is started before the thing it needs exists.
+
+```
+1  install and honesty      US-41  US-40  US-56  US-23  US-26  US-24
+                            US-57
+     a stranger can run it, and it tells the truth about what it read
+
+2  identity and persistence US-35  US-42  US-33  US-22
+     results stop being disposable, and a finding keeps its name
+
+3  scale and automation     US-21  US-43  US-44  US-20  US-46  US-47
+     it survives a real estate and runs without being asked
+
+4  breadth and content      US-27  US-28  US-53  US-34  US-54  US-31
+     more vendors, real benchmarks, and the fix as well as the fault
+
+5  the buyer's questions    US-50  US-51  US-52  US-45  US-37  US-48
+                            US-49
+     procurement, security review, and the team-sized product
+
+throughout                  US-25  US-29  US-30  US-32  US-36
+                            US-38  US-39  US-55
+     independent of the chain; can start any time
+```
+
+---
+
+## What would make this milestone a failure
+
+Worth writing down now, while it is cheap to say.
+
+1. **A story marked done that widened a claim rather than a capability.** US-24, US-40 and US-43 are all one careless commit away from asserting more coverage than they deliver. That is the exact failure this product exists to catch in other people's networks.
+2. **Reviews thinning as authoring concentrates.** See the section above. This is the most likely way it goes wrong.
+3. **This document going stale.** It is generated from a single source along with the issues themselves, precisely so it cannot disagree with them. If it is ever hand-edited, that guarantee is gone and it becomes one more copy of a fact — which is the defect family this project has been recording since the first sprint.
+
@@ -0,0 +1,100 @@
+"""`docs/design/phase-3-backlog.md` must be what the generator produces.
+
+WHY THIS FILE EXISTS
+    The document says, in its own last section, that it cannot disagree with
+    the GitHub issues because both are generated from one source. That claim
+    is only true while nobody hand-edits the document. Nothing enforced it,
+    which made it exactly the kind of sentence this project keeps writing and
+    then falsifying -- a described practice standing in for a performed one.
+
+    So the claim is checked instead of asserted.
+
+WHY THIS IS NOT #267
+    #267 is the reason to be careful here. `docs/traceability.md` carries
+    per-file test counts, so a CI check on it would be red every time anybody
+    adds a test -- permanently red, and therefore ignored.
+
+    This document carries no count that `main` can move. Its content is a
+    function of the story list and the recorded issue numbers, both of which
+    live in files in this repository. It changes when somebody edits the
+    generator, and at no other time. That is what makes the check stable
+    rather than noise.
+
+WHAT FAILURE MEANS
+    Either the document was edited by hand -- regenerate it and put the change
+    in the generator instead -- or the generator changed and the document was
+    not regenerated. Both are the same defect and both have the same fix:
+
+        python -m tools.make_phase3_backlog doc
+
+These need neither Batfish nor Ollama.
+"""
+
+from __future__ import annotations
+
+import importlib.util
+import sys
+from pathlib import Path
+
+import pytest
+
+REPO_ROOT = Path(__file__).resolve().parent.parent
+GENERATOR = REPO_ROOT / "tools" / "make_phase3_backlog.py"
+DOCUMENT = REPO_ROOT / "docs" / "design" / "phase-3-backlog.md"
+
+
+def _generator():
+    """Import the script by path -- tools/ is not a package on sys.path."""
+    spec = importlib.util.spec_from_file_location("_phase3", GENERATOR)
+    module = importlib.util.module_from_spec(spec)
+    sys.modules["_phase3"] = module
+    spec.loader.exec_module(module)
+    return module
+
+
+@pytest.fixture(scope="module")
+def gen():
+    if not GENERATOR.exists():          # pragma: no cover - the file is tracked
+        pytest.skip("generator not present")
+    return _generator()
+
+
+def test_document_matches_the_generator(gen):
+    """The committed document is byte-identical to a fresh generation."""
+    on_disk = DOCUMENT.read_text(encoding="utf-8")
+    assert on_disk == gen.document(), (
+        "docs/design/phase-3-backlog.md is not what the generator produces. "
+        "Run: python -m tools.make_phase3_backlog doc"
+    )
+
+
+def test_every_story_has_an_issue_number(gen):
+    """A backlog row with no issue behind it is a plan, not work."""
+    import json
+
+    numbers = json.loads(gen.NUMBERS.read_text(encoding="utf-8"))
+    every = {row[0] for row in gen.STORIES} | {e[0] for e in gen.EXISTING}
+    assert set(numbers) == every, (
+        f"stories with no issue: {sorted(every - set(numbers))}; "
+        f"issues with no story: {sorted(set(numbers) - every)}"
+    )
+
+
+def test_story_ids_are_unique(gen):
+    """Two stories sharing an id is the duplicate-`id` bug in a backlog."""
+    ids = [row[0] for row in gen.STORIES] + [e[0] for e in gen.EXISTING]
+    assert len(ids) == len(set(ids)), "duplicate story id"
+
+
+def test_every_story_is_assigned_to_a_real_member(gen):
+    """Nothing here is owned by nobody -- that is how a backlog rots."""
+    members = {gen.A, gen.N, gen.S, gen.K}
+    for row in gen.STORIES:
+        assert row[5] in members, f"{row[0]} assigned to {row[5]!r}"
+
+
+def test_every_story_states_why_it_is_in_the_milestone(gen):
+    """A story with no argument behind it is a wish, and gets cut."""
+    for row in gen.STORIES:
+        assert row[3], f"{row[0]} has no acceptance criteria"
+        assert len(row[4].split()) >= 12, f"{row[0]} has no real rationale"