backdraft 0.2.0__tar.gz → 0.2.1__tar.gz

{backdraft-0.2.0 → backdraft-0.2.1}/DESIGN.md

@@ -140,7 +140,7 @@ substrates); this list is questions still awaiting a decision:
 - Re-bind/orphan pass on re-ingest of changed docs (chunk ordinal drift)
 - Prose-file locator forms; slug assignment/collision rules (the trailing-hyphen
   truncation from the Darnell trial lives here)
-- Distribution: PyPI (name `backdraft` verified free 2026-07-28), skill
-  registration; domain backdraft.dev acquired 2026-07-28
+- Distribution: published to PyPI as `backdraft` 0.2.0 (2026-07-28); domain
+  backdraft.dev acquired; still open: skill registration, deploying site/
 - First external user (LoanCore validation layer vs Clarion eval substrate vs
   open release feedback)

{backdraft-0.2.0 → backdraft-0.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: backdraft
-Version: 0.2.0
+Version: 0.2.1
 Summary: Drop-in provenance for factual claims: one click from any claim to the evidence behind it.
 License: MIT
 License-File: LICENSE
@@ -27,24 +27,24 @@ or does not use the output. Backdraft removes that cost. In the usual loop, an
 **agent** does the writing: it reads your source documents only through a gate
 that mints a citation token over every span it shows, cites those tokens
 inline, and a postprocess resolves them, attaches the verbatim evidence, and
-renders one self-contained HTML file — the memo with the actual cited pages
+renders one self-contained HTML file, the memo with the actual cited pages
 and spreadsheet cells embedded inside it. The **human** just reads that file:
 click a claim, see the source.
 
 It works the same whether the writer is a model or a person, and whether the
 citations were minted during writing or attached to a document afterwards.
 
-Docs live at [backdraft.dev](https://backdraft.dev) — one page for people, and
+Docs live at [backdraft.dev](https://backdraft.dev), one page for people, and
 [backdraft.dev/llms.txt](https://backdraft.dev/llms.txt) for agents.
 
 ## 60 seconds
 
 ```bash
-uv tool install git+https://github.com/spencerbraun/backdraft
+uv tool install "backdraft[vlm]"
 cd demo
 ```
 
-**Ingest** the sources. Every anchor is minted here — chunks for PDF pages,
+**Ingest** the sources. Every anchor is minted here, chunks for PDF pages,
 cells for spreadsheet sheets.
 
 ```console
@@ -66,7 +66,7 @@ $ backdraft read t12-summary p1
 # t12-summary p1  (page 1 of 3)
 
 [bd:t12-summary:p1.c1:c2e8]
-Bridgeview Commons — Trailing Twelve Month Summary
+Bridgeview Commons, Trailing Twelve Month Summary
 
 Property: Bridgeview Commons, 4400 Halsted Avenue, Columbus, OH 43214. 128 units across four three-story
 garden buildings on 6.2 acres, built in 1998 and substantially renovated between 2019 and 2021. This summary
@@ -86,7 +86,7 @@ Chunks fall on the document's own paragraphs. A PDF stores glyphs at
 coordinates, not paragraphs, so the extractor rebuilds the breaks from the line
 geometry before the chunker sees the page.
 
-Search results are citable too — no page read required to get an anchor:
+Search results are citable too, no page read required to get an anchor:
 
 ```console
 $ backdraft search "24850000"
@@ -129,12 +129,12 @@ $ echo $?
 2
 ```
 
-Exit 2 means a citation did not resolve — the code a `Stop` hook or a CI job
+Exit 2 means a citation did not resolve, the code a `Stop` hook or a CI job
 gates on.
 
 **Render.** One self-contained HTML file: the document, the receipts, the
-evidence — cited page images, spreadsheet cells in place, the machine-readable
-record — all embedded, no network, nothing to install for the reader.
+evidence, cited page images, spreadsheet cells in place, the machine-readable
+record, all embedded, no network, nothing to install for the reader.
 
 ```console
 $ backdraft render memo.md --to html
@@ -152,13 +152,13 @@ in at [`demo/memo.backdraft.html`](demo/memo.backdraft.html).
 
 Reading is a **gate**: source documents reach the writer only through `read` and
 `search`, which mint a token over everything they show and record it in a
-**ledger**, so the set of citable things is exactly the set of things shown — and
+**ledger**, so the set of citable things is exactly the set of things shown, and
 a citation to something the writer never saw is a distinguishable failure rather
 than an invisible one. **Anchors are content-addressed**: a token's identity
 derives from the document's content and a location inside it, and it binds to an
 extraction snapshot, so re-ingesting the same bytes yields the same tokens and a
 changed source yields `drifted` instead of a broken link. **The receipt travels
-with the claim** — an anchor is not a pointer but a pointer plus the verbatim
+with the claim**, an anchor is not a pointer but a pointer plus the verbatim
 snippet and its hash, which is why the finished artifact is defensible with the
 registry deleted and the sources gone. **Verification is a set of switches,
 default off**: resolution is inherent to binding, but `value-trace`, `overlap`
@@ -175,31 +175,29 @@ artifact. Nothing is ever warned about and dropped.
 ## What lives where
 
 A working directory stays clean: the authored document and its artifact are
-the only visible outputs. Everything else — registry, credentials, bind
-records — lives under one hidden `.backdraft/` directory, and
+the only visible outputs. Everything else, registry, credentials, bind
+records, lives under one hidden `.backdraft/` directory, and
 `backdraft clean` tidies strays from older versions. The registry contains the
 full text of everything ingested, so **gitignore `.backdraft/` for
 confidential corpora**.
 
 ## Install
 
-Not yet on PyPI — install from the repository:
-
 ```bash
-uv tool install git+https://github.com/spencerbraun/backdraft      # CLI
-uv add "backdraft @ git+https://github.com/spencerbraun/backdraft" # as a dependency
-pip install git+https://github.com/spencerbraun/backdraft          # or plain pip
+uv tool install "backdraft[vlm]"   # CLI, recommended form
+uv add backdraft                   # as a dependency
+pip install backdraft              # or plain pip
 ```
 
 Python 3.13+. Two extras:
 
 | Extra | Adds | For |
 |---|---|---|
-| `[vlm]` | `openai`, `pdf2image` | the VLM extractor — the recommended path for real PDFs (glossy layouts, info boxes, scans): a vision model reads each page and its clean representation becomes the receipt. `--extractor auto` prefers it only on explicit, backdraft-scoped consent: `BACKDRAFT_VLM_API_KEY` (env or `.backdraft/env` — `backdraft init` writes a template). Ambient `OPENAI_API_KEY`/`OPENROUTER_API_KEY` are deliberately never read; a generic key in the environment is not consent to send documents to its provider. The default provider is OpenRouter running Gemini 3.1 Flash Lite; direct OpenAI is base_url + model, set explicitly. Without a scoped key, `auto` falls back to the keyless text layer and says so. Needs poppler installed separately |
+| `[vlm]` | `openai`, `pdf2image` | the VLM extractor, the recommended path for real PDFs (glossy layouts, info boxes, scans): a vision model reads each page and its clean representation becomes the receipt. `--extractor auto` prefers it only on explicit, backdraft-scoped consent: `BACKDRAFT_VLM_API_KEY` (env or `.backdraft/env`, `backdraft init` writes a template). Ambient `OPENAI_API_KEY`/`OPENROUTER_API_KEY` are deliberately never read; a generic key in the environment is not consent to send documents to its provider. The default provider is OpenRouter running Gemini 3.1 Flash Lite; direct OpenAI is base_url + model, set explicitly. Without a scoped key, `auto` falls back to the keyless text layer and says so. Needs poppler installed separately |
 | `[entail]` | `anthropic` | `bind --check entail`, the model-judge verifier |
 
 ```bash
-pip install "backdraft[vlm,entail] @ git+https://github.com/spencerbraun/backdraft"
+pip install "backdraft[vlm,entail]"
 ```
 
 From a checkout:
@@ -213,7 +211,7 @@ uv run pytest
 ## Skills
 
 Three agent skills ship in [`skills/`](skills). Copy the ones you want into your
-agent's skills directory — for Claude Code, `~/.claude/skills/` (personal) or
+agent's skills directory, for Claude Code, `~/.claude/skills/` (personal) or
 `.claude/skills/` (per project):
 
 ```bash
@@ -233,11 +231,11 @@ the skills is required to use backdraft by hand.
 
 | Document | Is |
 |---|---|
-| [DESIGN.md](DESIGN.md) | why it is shaped this way — principles, architecture, decision log |
-| [SPEC.md](SPEC.md) | the builders' contract — types, grammar, DDL, CLI surface, module boundaries |
+| [DESIGN.md](DESIGN.md) | why it is shaped this way, principles, architecture, decision log |
+| [SPEC.md](SPEC.md) | the builders' contract, types, grammar, DDL, CLI surface, module boundaries |
 | [spec/tokens.md](spec/tokens.md) | the citation token grammar, normatively |
 | [spec/chunking.md](spec/chunking.md) | the deterministic chunker |
-| [spec/artifact.md](spec/artifact.md) | the artifact format — sidecar payload, legend, HTML rules |
+| [spec/artifact.md](spec/artifact.md) | the artifact format, sidecar payload, legend, HTML rules |
 | [demo/walkthrough.md](demo/walkthrough.md) | the whole thing end to end, with real output |
 
 The three files under `spec/` are the portable specification: another
@@ -246,8 +244,8 @@ implementation reads them and nothing else.
 ## Status
 
 v1. The registry format, the token grammar and the artifact format are pinned
-by prose specs and golden-file tests. What is deliberately queued — exhibits,
-Excel region maps, more formats, `bd:calc` derivations — is written down in
+by prose specs and golden-file tests. What is deliberately queued, exhibits,
+Excel region maps, more formats, `bd:calc` derivations, is written down in
 [ROADMAP.md](ROADMAP.md).
 
 MIT licensed.

{backdraft-0.2.0 → backdraft-0.2.1}/README.md

@@ -10,24 +10,24 @@ or does not use the output. Backdraft removes that cost. In the usual loop, an
 **agent** does the writing: it reads your source documents only through a gate
 that mints a citation token over every span it shows, cites those tokens
 inline, and a postprocess resolves them, attaches the verbatim evidence, and
-renders one self-contained HTML file — the memo with the actual cited pages
+renders one self-contained HTML file, the memo with the actual cited pages
 and spreadsheet cells embedded inside it. The **human** just reads that file:
 click a claim, see the source.
 
 It works the same whether the writer is a model or a person, and whether the
 citations were minted during writing or attached to a document afterwards.
 
-Docs live at [backdraft.dev](https://backdraft.dev) — one page for people, and
+Docs live at [backdraft.dev](https://backdraft.dev), one page for people, and
 [backdraft.dev/llms.txt](https://backdraft.dev/llms.txt) for agents.
 
 ## 60 seconds
 
 ```bash
-uv tool install git+https://github.com/spencerbraun/backdraft
+uv tool install "backdraft[vlm]"
 cd demo
 ```
 
-**Ingest** the sources. Every anchor is minted here — chunks for PDF pages,
+**Ingest** the sources. Every anchor is minted here, chunks for PDF pages,
 cells for spreadsheet sheets.
 
 ```console
@@ -49,7 +49,7 @@ $ backdraft read t12-summary p1
 # t12-summary p1  (page 1 of 3)
 
 [bd:t12-summary:p1.c1:c2e8]
-Bridgeview Commons — Trailing Twelve Month Summary
+Bridgeview Commons, Trailing Twelve Month Summary
 
 Property: Bridgeview Commons, 4400 Halsted Avenue, Columbus, OH 43214. 128 units across four three-story
 garden buildings on 6.2 acres, built in 1998 and substantially renovated between 2019 and 2021. This summary
@@ -69,7 +69,7 @@ Chunks fall on the document's own paragraphs. A PDF stores glyphs at
 coordinates, not paragraphs, so the extractor rebuilds the breaks from the line
 geometry before the chunker sees the page.
 
-Search results are citable too — no page read required to get an anchor:
+Search results are citable too, no page read required to get an anchor:
 
 ```console
 $ backdraft search "24850000"
@@ -112,12 +112,12 @@ $ echo $?
 2
 ```
 
-Exit 2 means a citation did not resolve — the code a `Stop` hook or a CI job
+Exit 2 means a citation did not resolve, the code a `Stop` hook or a CI job
 gates on.
 
 **Render.** One self-contained HTML file: the document, the receipts, the
-evidence — cited page images, spreadsheet cells in place, the machine-readable
-record — all embedded, no network, nothing to install for the reader.
+evidence, cited page images, spreadsheet cells in place, the machine-readable
+record, all embedded, no network, nothing to install for the reader.
 
 ```console
 $ backdraft render memo.md --to html
@@ -135,13 +135,13 @@ in at [`demo/memo.backdraft.html`](demo/memo.backdraft.html).
 
 Reading is a **gate**: source documents reach the writer only through `read` and
 `search`, which mint a token over everything they show and record it in a
-**ledger**, so the set of citable things is exactly the set of things shown — and
+**ledger**, so the set of citable things is exactly the set of things shown, and
 a citation to something the writer never saw is a distinguishable failure rather
 than an invisible one. **Anchors are content-addressed**: a token's identity
 derives from the document's content and a location inside it, and it binds to an
 extraction snapshot, so re-ingesting the same bytes yields the same tokens and a
 changed source yields `drifted` instead of a broken link. **The receipt travels
-with the claim** — an anchor is not a pointer but a pointer plus the verbatim
+with the claim**, an anchor is not a pointer but a pointer plus the verbatim
 snippet and its hash, which is why the finished artifact is defensible with the
 registry deleted and the sources gone. **Verification is a set of switches,
 default off**: resolution is inherent to binding, but `value-trace`, `overlap`
@@ -158,31 +158,29 @@ artifact. Nothing is ever warned about and dropped.
 ## What lives where
 
 A working directory stays clean: the authored document and its artifact are
-the only visible outputs. Everything else — registry, credentials, bind
-records — lives under one hidden `.backdraft/` directory, and
+the only visible outputs. Everything else, registry, credentials, bind
+records, lives under one hidden `.backdraft/` directory, and
 `backdraft clean` tidies strays from older versions. The registry contains the
 full text of everything ingested, so **gitignore `.backdraft/` for
 confidential corpora**.
 
 ## Install
 
-Not yet on PyPI — install from the repository:
-
 ```bash
-uv tool install git+https://github.com/spencerbraun/backdraft      # CLI
-uv add "backdraft @ git+https://github.com/spencerbraun/backdraft" # as a dependency
-pip install git+https://github.com/spencerbraun/backdraft          # or plain pip
+uv tool install "backdraft[vlm]"   # CLI, recommended form
+uv add backdraft                   # as a dependency
+pip install backdraft              # or plain pip
 ```
 
 Python 3.13+. Two extras:
 
 | Extra | Adds | For |
 |---|---|---|
-| `[vlm]` | `openai`, `pdf2image` | the VLM extractor — the recommended path for real PDFs (glossy layouts, info boxes, scans): a vision model reads each page and its clean representation becomes the receipt. `--extractor auto` prefers it only on explicit, backdraft-scoped consent: `BACKDRAFT_VLM_API_KEY` (env or `.backdraft/env` — `backdraft init` writes a template). Ambient `OPENAI_API_KEY`/`OPENROUTER_API_KEY` are deliberately never read; a generic key in the environment is not consent to send documents to its provider. The default provider is OpenRouter running Gemini 3.1 Flash Lite; direct OpenAI is base_url + model, set explicitly. Without a scoped key, `auto` falls back to the keyless text layer and says so. Needs poppler installed separately |
+| `[vlm]` | `openai`, `pdf2image` | the VLM extractor, the recommended path for real PDFs (glossy layouts, info boxes, scans): a vision model reads each page and its clean representation becomes the receipt. `--extractor auto` prefers it only on explicit, backdraft-scoped consent: `BACKDRAFT_VLM_API_KEY` (env or `.backdraft/env`, `backdraft init` writes a template). Ambient `OPENAI_API_KEY`/`OPENROUTER_API_KEY` are deliberately never read; a generic key in the environment is not consent to send documents to its provider. The default provider is OpenRouter running Gemini 3.1 Flash Lite; direct OpenAI is base_url + model, set explicitly. Without a scoped key, `auto` falls back to the keyless text layer and says so. Needs poppler installed separately |
 | `[entail]` | `anthropic` | `bind --check entail`, the model-judge verifier |
 
 ```bash
-pip install "backdraft[vlm,entail] @ git+https://github.com/spencerbraun/backdraft"
+pip install "backdraft[vlm,entail]"
 ```
 
 From a checkout:
@@ -196,7 +194,7 @@ uv run pytest
 ## Skills
 
 Three agent skills ship in [`skills/`](skills). Copy the ones you want into your
-agent's skills directory — for Claude Code, `~/.claude/skills/` (personal) or
+agent's skills directory, for Claude Code, `~/.claude/skills/` (personal) or
 `.claude/skills/` (per project):
 
 ```bash
@@ -216,11 +214,11 @@ the skills is required to use backdraft by hand.
 
 | Document | Is |
 |---|---|
-| [DESIGN.md](DESIGN.md) | why it is shaped this way — principles, architecture, decision log |
-| [SPEC.md](SPEC.md) | the builders' contract — types, grammar, DDL, CLI surface, module boundaries |
+| [DESIGN.md](DESIGN.md) | why it is shaped this way, principles, architecture, decision log |
+| [SPEC.md](SPEC.md) | the builders' contract, types, grammar, DDL, CLI surface, module boundaries |
 | [spec/tokens.md](spec/tokens.md) | the citation token grammar, normatively |
 | [spec/chunking.md](spec/chunking.md) | the deterministic chunker |
-| [spec/artifact.md](spec/artifact.md) | the artifact format — sidecar payload, legend, HTML rules |
+| [spec/artifact.md](spec/artifact.md) | the artifact format, sidecar payload, legend, HTML rules |
 | [demo/walkthrough.md](demo/walkthrough.md) | the whole thing end to end, with real output |
 
 The three files under `spec/` are the portable specification: another
@@ -229,8 +227,8 @@ implementation reads them and nothing else.
 ## Status
 
 v1. The registry format, the token grammar and the artifact format are pinned
-by prose specs and golden-file tests. What is deliberately queued — exhibits,
-Excel region maps, more formats, `bd:calc` derivations — is written down in
+by prose specs and golden-file tests. What is deliberately queued, exhibits,
+Excel region maps, more formats, `bd:calc` derivations, is written down in
 [ROADMAP.md](ROADMAP.md).
 
 MIT licensed.

{backdraft-0.2.0 → backdraft-0.2.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "backdraft"
-version = "0.2.0"
+version = "0.2.1"
 description = "Drop-in provenance for factual claims: one click from any claim to the evidence behind it."
 readme = "README.md"
 requires-python = ">=3.13"
@@ -36,6 +36,9 @@ build-backend = "hatchling.build"
 [tool.hatch.build.targets.wheel]
 packages = ["src/backdraft"]
 
+[tool.hatch.build.targets.wheel.force-include]
+"skills" = "backdraft/skills"
+
 [tool.pytest.ini_options]
 testpaths = ["tests"]
 addopts = "-q"

{backdraft-0.2.0 → backdraft-0.2.1}/site/docs.html

@@ -3,7 +3,7 @@
 <head>
 <meta charset="utf-8">
 <meta name="viewport" content="width=device-width, initial-scale=1">
-<title>Backdraft — documentation</title>
+<title>Backdraft, documentation</title>
 <meta name="description" content="Install, concepts, the full workflow, CLI reference, verification, the artifact format, and agent setup for Backdraft.">
 <link rel="icon" href='data:image/svg+xml,%3Csvg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 64 64"%3E%3Crect width="64" height="64" rx="13" fill="%23FBFAF6"/%3E%3Cg transform="translate(32 32) scale(0.82) translate(-32 -32)"%3E%3Cpath fill="%23282828" fill-rule="evenodd" d="M44 3 C38 8 30 10 26 16 C23.5 20 24.5 24 27 26.5 C20.5 28 14.5 33 14 41 C13.3 51.5 22 59.5 33 59.5 C44 59.5 51.5 51.5 51 41.5 C50.6 33.5 45 29 42.5 22.5 C40.8 18 42.5 9 44 3 Z M34 34 C38 38.5 39.5 43.5 37.5 47.5 C35 52 28.5 51.5 26.5 47 C25 43.5 27.5 40 30 38.5 C31.8 37.5 33.2 36 34 34 Z"/%3E%3C/g%3E%3C/svg%3E'>
 <style>
@@ -106,25 +106,37 @@ footer{text-align:center;font-family:var(--sans);font-size:.74rem;color:var(--mu
 
 <main>
 <h1>Documentation</h1>
-<p class="lede">Everything to run the loop: install, the concepts that make it
-trustworthy, every command, and what ends up on disk. The normative format
-specs live <a href="https://github.com/spencerbraun/backdraft/tree/main/spec">in
-the repo</a>; agents should start from <a href="llms.txt">llms.txt</a>.</p>
+<p class="lede">Backdraft is three things that ship together: a Python CLI
+(<code>pip install backdraft</code>) that ingests documents and checks
+citations, a one-page agent skill that makes your agent write through it, and
+a self-contained file format for the output. There is no server and no SDK to
+integrate, the CLI is the whole system. This page covers install, the
+concepts, every command, and what ends up on disk. The normative format specs
+live <a href="https://github.com/spencerbraun/backdraft/tree/main/spec">in the
+repo</a>, and agents should start from <a href="llms.txt">llms.txt</a>.</p>
 
 <section id="install">
 <h2>Install</h2>
-<pre>uv tool install "backdraft[vlm] @ git+https://github.com/spencerbraun/backdraft"</pre>
+<pre>uv tool install "backdraft[vlm]"</pre>
 <p>Python 3.13+. The <code>[vlm]</code> extra adds the vision-model PDF
-extractor — the recommended path for real PDFs (glossy layouts, info boxes,
-scans) — and needs <code>poppler</code> on your PATH
+extractor, the recommended path for real PDFs (glossy layouts, info boxes,
+scans), and needs <code>poppler</code> on your PATH
 (<code>brew install poppler</code>). Without it, PDFs fall back to the keyless
 embedded-text layer, and ingest says so.</p>
-<p>The VLM extractor runs only on explicit, backdraft-scoped consent: set
-<code>BACKDRAFT_VLM_API_KEY</code> in <code>.backdraft/env</code> (written by
-<code>backdraft init</code>) or the environment. The default provider is
-OpenRouter running Gemini Flash; ambient <code>OPENAI_API_KEY</code>-style
-variables are deliberately never read. <code>[entail]</code> adds the optional
-model-judge verifier.</p>
+<h3>Bring a model provider</h3>
+<p>The vision extractor needs a key you provide, and it runs only on explicit,
+backdraft-scoped consent: set <code>BACKDRAFT_VLM_API_KEY</code> in
+<code>.backdraft/env</code> (written by <code>backdraft init</code>) or the
+environment. Ambient <code>OPENAI_API_KEY</code>-style variables are
+deliberately never read.</p>
+<p>Under the hood the client is the OpenAI SDK with an injectable
+<code>base_url</code>, so any OpenAI-compatible provider works. The default is
+OpenRouter (<code>https://openrouter.ai/api/v1</code>) running
+<code>google/gemini-3.1-flash-lite-preview</code>, so the simplest setup is an
+OpenRouter key. To point elsewhere, set <code>BACKDRAFT_VLM_BASE_URL</code>
+and <code>BACKDRAFT_VLM_MODEL</code>, for example directly at OpenAI or at a
+local server. <code>[entail]</code> adds the optional model-judge
+verifier, keyed separately by <code>BACKDRAFT_ENTAIL_API_KEY</code>.</p>
 </section>
 
 <section id="concepts">
@@ -133,11 +145,11 @@ model-judge verifier.</p>
 <p>Source documents reach the writer only through <code>read</code>,
 <code>search</code> and <code>cell</code>, which stamp a <b>token</b> on every
 span they show and record the showing in a session <b>ledger</b>. The set of
-citable things is exactly the set of things shown — so a citation to something
+citable things is exactly the set of things shown, so a citation to something
 the writer never saw is a distinguishable failure
 (<code>not_shown</code>), not an invisible one.</p>
 <h3>Tokens and receipts</h3>
-<p><code>bd:&lt;slug&gt;:&lt;locator&gt;:&lt;hash&gt;</code> — document,
+<p><code>bd:&lt;slug&gt;:&lt;locator&gt;:&lt;hash&gt;</code>: document,
 place, and a content-hash of the exact text cited. Locators are media-native:
 <code>p8</code> (a page), <code>p8.c3</code> (a paragraph-scale chunk),
 <code>rent-roll!B10</code> (a cell). An anchor is not a pointer: it carries the
@@ -146,7 +158,7 @@ the registry deleted and the sources gone.</p>
 <h3>Drift</h3>
 <p>Re-ingesting identical bytes yields identical tokens. When a source
 changes, citations into it resolve against the superseded snapshot and report
-<code>drifted</code> — the artifact shows what was cited and what stands now,
+<code>drifted</code>. The artifact shows what was cited and what stands now,
 as a word-diff.</p>
 </section>
 
@@ -196,7 +208,7 @@ skill</a> imposes the gate, and you receive the artifact.</div>
 <tr><th>Code</th><th>Means</th></tr>
 <tr><td><code>0</code></td><td>Every citation resolved.</td></tr>
 <tr><td><code>1</code></td><td>Usage or environment error.</td></tr>
-<tr><td><code>2</code></td><td>Bind completed and something did not resolve — the code a CI job or Stop-hook gates on. Verification verdicts never produce it.</td></tr>
+<tr><td><code>2</code></td><td>Bind completed and something did not resolve, the code a CI job or Stop-hook gates on. Verification verdicts never produce it.</td></tr>
 </table>
 </section>
 
@@ -211,7 +223,7 @@ output; multiple citations are <code>;</code>-separated in one href:</p>
 evidence supports.</li>
 <li><b>Cite only what you were shown.</b> Never construct or edit a token by
 hand.</li>
-<li><b>Uncited prose is fine</b> for recommendations and framing — cite facts,
+<li><b>Uncited prose is fine</b> for recommendations and framing, cite facts,
 not opinions.</li>
 <li><b>Never fix a failure by deleting its token.</b> A kept failure is the
 honest outcome; the artifact will show it plainly.</li>
@@ -222,7 +234,7 @@ artifact's subtitle.</li>
 
 <section id="verification">
 <h2>Verification</h2>
-<p>Independent switches, off by default, recorded as graded evidence — never
+<p>Independent switches, off by default, recorded as graded evidence, never
 gates. <code>value-trace</code> finds every figure in a claim in its cited
 source, reading through thousands separators, currency, scale suffixes
 ($1.4M ≡ 1,400,000), percent-vs-decimal (7.7% ≡ 0.077), accounting negatives
@@ -238,11 +250,11 @@ Record layer in plain language.</p>
 <h2>Files &amp; the artifact</h2>
 <table>
 <tr><th>File</th><th>Is</th></tr>
-<tr><td><code>memo.md</code></td><td>What the writer wrote — prose plus tokens. Yours; re-bindable.</td></tr>
+<tr><td><code>memo.md</code></td><td>What the writer wrote, prose plus tokens. Yours; re-bindable.</td></tr>
 <tr><td><code>memo.backdraft.html</code></td><td>The deliverable: document, receipts and evidence in one self-contained file. No network (CSP-enforced), nothing to install, degrades to readable footnotes if scripts are stripped.</td></tr>
-<tr><td><code>.backdraft/</code></td><td>The machinery: registry, credentials, and the bind record (<code>records/memo.backdraft.json</code> — the machine-readable run, also embedded in the artifact as a JSON island with a self-describing legend).</td></tr>
+<tr><td><code>.backdraft/</code></td><td>The machinery: registry, credentials, and the bind record (<code>records/memo.backdraft.json</code>, the machine-readable run, also embedded in the artifact as a JSON island with a self-describing legend).</td></tr>
 </table>
-<p>The artifact embeds only <em>cited</em> evidence — a memo citing ten pages
+<p>The artifact embeds only <em>cited</em> evidence, a memo citing ten pages
 of a gigabyte corpus is a ~2&nbsp;MB file. Recipients need the one
 <code>.html</code> file and nothing else.</p>
 </section>
@@ -266,18 +278,18 @@ token-efficient reference for any agent is <a href="llms.txt">llms.txt</a>
 <code>--config</code>. A generic <code>OPENAI_API_KEY</code> in your shell is
 not consent to send documents anywhere.</li>
 <li><b>Artifacts make no network requests</b>, enforced by a
-Content-Security-Policy the browser applies — no fonts, no analytics, no
+Content-Security-Policy the browser applies, no fonts, no analytics, no
 phone-home.</li>
 <li><b>Everything is local.</b> The only network calls in the system are the
 optional VLM/entail model calls you explicitly key.</li>
-<li><b>Gitignore <code>.backdraft/</code></b> for confidential corpora — the
+<li><b>Gitignore <code>.backdraft/</code></b> for confidential corpora, the
 registry contains the full text of everything ingested.</li>
 </ul>
 </section>
 
 <section id="faq">
 <h2>FAQ</h2>
-<h3>Ingest says it fell back to the text layer — why?</h3>
+<h3>Ingest says it fell back to the text layer, why?</h3>
 <p>The note names the condition: either the <code>[vlm]</code> extra isn't
 installed, or <code>BACKDRAFT_VLM_API_KEY</code> isn't set. Fix the named one.
 Text-layer receipts are fine for clean digital PDFs; glossy or scanned ones
@@ -288,7 +300,7 @@ registries, run <code>backdraft snapshot-pages &lt;slug&gt;</code> (local,
 free), then re-bind.</p>
 <h3>What does <code>not_shown</code> actually catch?</h3>
 <p>A real token, valid in the registry, that the writing session was never
-shown — i.e. the writer cited something it didn't read. Only a session ledger
+shown, i.e. the writer cited something it didn't read. Only a session ledger
 makes this class of failure visible; that is why the skill starts one.</p>
 <h3>Can a reader trust an artifact without installing backdraft?</h3>
 <p>That's the design goal: the receipts, the evidence, and a machine-readable
@@ -296,7 +308,7 @@ record with its own decoding legend are inside the file. The
 <a href="https://github.com/spencerbraun/backdraft/blob/main/spec/artifact.md">artifact
 spec</a> lists the checks a skeptic can run with nothing but the file.</p>
 <h3>Is this fact-checking?</h3>
-<p>No — provenance. Backdraft proves <i>where a claim came from</i> and shows
+<p>No, provenance. Backdraft proves <i>where a claim came from</i> and shows
 you the source; the optional verifiers add deterministic evidence like "this
 figure appears in that cell." Judgment stays with the reader.</p>
 </section>