tolkira 0.0.0-stage → 0.1.1

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.
package/DESIGN.md ADDED
@@ -0,0 +1,178 @@
1
+ # Tolkira report design system
2
+
3
+ **Mode:** Operate. The report is a work surface for developers and QA, not a
4
+ landing page. The tool disappears into the task: find the failure, read the
5
+ error, inspect the screenshot, move on.
6
+
7
+ ## Product job
8
+
9
+ Answer five questions in under thirty seconds:
10
+
11
+ 1. Did everything pass?
12
+ 2. What failed, and why?
13
+ 3. What did it look like?
14
+ 4. Can I find the relevant scenario?
15
+ 5. Can I share one file with someone else?
16
+
17
+ The default report is simple and failed-first. Deeper analysis belongs in the
18
+ optional Details view, never in the way of the failure.
19
+
20
+ ## Principles
21
+
22
+ - **One file, no network.** All CSS, scripts and attachments are inline. No
23
+ CDN, no remote font, no telemetry.
24
+ - **Status is text, not only colour.** Every status has a label; colour is a
25
+ second signal.
26
+ - **Evidence first.** A screenshot or attachment is part of the failure, not
27
+ decoration.
28
+ - **Native disclosure.** Use `<details>`/`<summary>` with a visible chevron;
29
+ search and filters open the features that contain matches.
30
+ - **Density with air.** Compact rows, but never cramped; 0.9375 rem body,
31
+ 0.75–0.875 rem metadata, 0.6875 rem minimum for functional text.
32
+ - **Accessible by default.** WCAG 2.2 AA, keyboard focus, reduced motion,
33
+ 200 % page zoom and 200 % text zoom, no colour-only status, no horizontal
34
+ page scroll.
35
+
36
+ ## Tokens
37
+
38
+ The source of truth is `src/render/styles.ts` (generated from the report CSS).
39
+ Use the token names, never raw colours.
40
+
41
+ | Token | Role |
42
+ | --- | --- |
43
+ | `--paper` | Page surface |
44
+ | `--canvas` | Page background |
45
+ | `--surface-2` | Secondary surface, row hover, panels |
46
+ | `--ink` | Primary text |
47
+ | `--muted` | Secondary text, labels, captions |
48
+ | `--line`, `--line-strong` | Hairline rules and control borders |
49
+ | `--accent`, `--accent-soft` | Brand, current selection, primary action |
50
+ | `--pass`, `--pass-soft` | Passed |
51
+ | `--fail`, `--fail-soft` | Failed |
52
+ | `--pending`, `--pending-soft` | Pending |
53
+ | `--undefined`, `--undefined-soft` | Undefined / missing step definition |
54
+ | `--skipped`, `--skipped-soft` | Skipped |
55
+ | `--radius`, `--shadow` | Panel radius and one elevation |
56
+
57
+ Dark mode is a token override under `prefers-color-scheme: dark`; it never
58
+ changes layout.
59
+
60
+ ## Type
61
+
62
+ - System sans for body and display; system mono only for measurements,
63
+ durations, file paths and error text.
64
+ - Body 0.9375 rem, line-height 1.5–1.55.
65
+ - Row title 0.8125–0.875 rem/600; metadata 0.6875–0.75 rem; step text
66
+ 0.75–0.8125 rem.
67
+ - Sentence case everywhere. No tracked capitals, no emoji icons.
68
+ - No fluid display headings in the report shell; fixed rem steps.
69
+
70
+ ### Why rem, not fluid type
71
+
72
+ `rem` ties every text size to the root font size, so the browser's
73
+ font-size setting and text-only zoom actually work. Fluid `clamp()`/`vw`
74
+ typography is for marketing pages: in a dense report it makes text small on
75
+ phones and unpredictable at 200 % zoom.
76
+
77
+ Fixed `px` stays only where it is not text: 1 px hairlines, focus-ring
78
+ thickness, radii and the 1320 px page frame. Spacing, control heights, icon
79
+ size, thumbnails and the fixed thumbnail gutter use `rem` so they grow with
80
+ the text they sit next to. Media-query breakpoints stay in `px` because they
81
+ describe the viewport, not the font.
82
+
83
+ ## Responsive
84
+
85
+ - Layout responds through content-driven primitives: `minmax()`, `flex-wrap`,
86
+ `max-width`, two breakpoints (900 px, 560 px) and container queries at
87
+ `16rem`. No fixed page width.
88
+ - The 720 × 450 browser test is **200 % page zoom** at a 1440 × 900 window, not
89
+ a hardcoded layout size. 320 px and 390 px are separate phone checks.
90
+ - Text zoom is tested by doubling the root font size; the layout must reflow
91
+ without horizontal page scroll. At `16rem` container width the feature and
92
+ scenario rows stack and the run title steps down, so text zoom keeps a
93
+ readable title column instead of squeezing it to zero.
94
+ - The shell is a document, not an app canvas: it has no viewport-height frame
95
+ and no horizontal scroll region except tables and code blocks. Every
96
+ horizontal scroll region is focusable (`tabindex="0"`, `role="region"`,
97
+ accessible name) so keyboard users can reach clipped columns.
98
+
99
+ ## Components
100
+
101
+ | Component | File / selector | Rule |
102
+ | --- | --- | --- |
103
+ | Top bar / tabs | `.topbar`, `.view` | Report and Details; active tab is a text state, not a card |
104
+ | Run header | `.runHead`, `.health` | Title, metadata, pass rate, segmented status bar |
105
+ | Toolbar | `.toolbar`, `.search`, `.filter` | Search and status filters; search opens matching features and scenarios; filters only for present statuses |
106
+ | Feature | `details.feature` | Native disclosure; status, name, path, counts, duration |
107
+ | Scenario | `details.scenario` | Fixed thumbnail gutter, status, name, tags, steps, duration |
108
+ | Step | `.step` | Status glyph plus screen-reader label, keyword, text, error, doc string, focusable table, attachments |
109
+ | Attachment | `.shot`, `.fileText`, `.fileLink` | Images inline; text in `<pre>`; binary as download; alt names the scenario and step |
110
+ | Details | `.analysis` | Feature table, status distribution, Run / Cucumber environment / Custom metadata blocks, agent summary |
111
+
112
+ ## Interaction rules
113
+
114
+ - Feature and scenario summaries are the only click targets that toggle; a
115
+ click inside the detail does not close it.
116
+ - Chevron rotates on open; `prefers-reduced-motion` removes the transition.
117
+ - Search and filters: hide non-matches, open features and matched scenarios,
118
+ update a polite live region, show an empty state when nothing matches.
119
+ - Switching tabs scrolls to the top; the previous panel's scroll position is
120
+ not restored.
121
+ - The lightbox locks background scrolling, restores focus to its trigger on
122
+ Escape, Close, backdrop click and padding click, and reserves header space so
123
+ Close never covers the screenshot.
124
+ - Status is never colour-only: scenario badges carry text, step markers carry a
125
+ shape glyph plus a screen-reader label, and forced-colours mode keeps the
126
+ glyph, marker border and disclosure chevron visible.
127
+ - The failure is open by default; everything else is scannable when collapsed.
128
+ - No right-click-only actions; use visible buttons or a `…` menu. Keyboard and
129
+ touch must reach the same actions.
130
+ - No modal for a task that inline disclosure can solve.
131
+
132
+ ## Styling policy
133
+
134
+ **Plain CSS with tokens, not Tailwind.** The report is generated as one HTML
135
+ string with inline CSS; Tailwind would add a build step, content scanning of
136
+ TypeScript template strings, and safelists for dynamic status classes. The
137
+ report has ~20 components and a 15 KB stylesheet. A small, explicit token
138
+ system is easier for agents to read, test and change than utility soup in
139
+ template strings. A local utility layer is fine when it removes repetition.
140
+
141
+ ## Agent workflow
142
+
143
+ 1. Read `docs/PRODUCT.md`, this file, `src/report/model.ts`,
144
+ `src/render/html.ts` and `src/render/styles.ts`.
145
+ 2. Change the view model first if the data is missing; keep `src/summary.ts`
146
+ independent.
147
+ 3. Use tokens and existing component classes; do not add a UI framework.
148
+ 4. Add a unit test for the model/HTML and a browser test for the interaction.
149
+ 5. Render a real fixture at 1440×900 and 390×844; check focus, overflow,
150
+ reduced motion, 200 % text zoom and dark mode.
151
+ 6. Run `npm run check` and `npm run test:e2e`.
152
+ 7. For a new visual world, use the Hallmark/Impeccable loop in
153
+ `docs/ui-loop/`; for a normal change, stay inside this system.
154
+
155
+ ## Current status
156
+
157
+ Direction A is implemented in production:
158
+
159
+ - warm paper tokens, hairline sections and monospace measurements;
160
+ - fixed thumbnail gutter on every scenario row, also on mobile;
161
+ - chevrons, failed-first disclosures and search that opens matching features;
162
+ - `undefined` is a distinct status, not a neutral one;
163
+ - Details tab with producer, runtime, OS, CPU, protocol and user metadata;
164
+ - empty state for search/filter results;
165
+ - screenshot lightbox with a close button and Escape support.
166
+
167
+ CSS lives in `src/render/report.css`; `scripts/build-styles.mjs` generates
168
+ `src/render/styles.ts` for the inline report.
169
+
170
+ Still open: a 200 % text-zoom pass with real content, and charts/trends in
171
+ Advanced.
172
+
173
+ ## Non-goals
174
+
175
+ - React, Tailwind, Bootstrap, jQuery, CDN fonts or charts.
176
+ - Old reporter themes.
177
+ - A server, watch mode, telemetry or test-management workflows.
178
+ - Right-click-only context menus.
package/GLOSSARY.md ADDED
@@ -0,0 +1,60 @@
1
+ # Glossary
2
+
3
+ **Tolkira** — the project and CLI. Coined from Scandinavian *tolk*
4
+ (interpreter).
5
+
6
+ **Report** — one self-contained HTML file generated from one or more test
7
+ result inputs.
8
+
9
+ **Legacy JSON** — the older Cucumber JSON output format. Implementation-specific
10
+ and lossy, but still widely used.
11
+
12
+ **Messages / Envelopes** — Cucumber Messages, the modern NDJSON envelope stream
13
+ used by official Cucumber implementations.
14
+
15
+ **Attachment** — a binary or text artefact attached to a test step, scenario or
16
+ hook. Screenshots are attachments.
17
+
18
+ **Embedding** — the legacy JSON representation of an attachment (`embeddings[]`).
19
+
20
+ **Adapter** — code that turns an input format into Cucumber Messages.
21
+
22
+ **Normalisation** — converting legacy JSON into Messages so one renderer can
23
+ handle both.
24
+
25
+ **Renderer** — Tolkira's server-side report renderer: it turns the report model
26
+ into HTML, CSS and a small script. It is part of Tolkira, not an upstream React
27
+ component tree.
28
+
29
+ **Shell** — Tolkira's own wrapper around the renderer: title, metadata, summary,
30
+ theme, logo.
31
+
32
+ **Summary** — the stable, renderer-independent JSON document emitted by
33
+ `--json` / `summarizeReport` for agents and CI. It is versioned and is not the
34
+ HTML report.
35
+
36
+ **Schema version** — the integer in `summary.schemaVersion`. A breaking change
37
+ increments it; consumers may reject versions they do not know.
38
+
39
+ **Fixture** — a small, real or realistic input file used by tests.
40
+
41
+ **Vertical slice** — a small end-to-end change: input → normalise → render →
42
+ output, with tests.
43
+
44
+ **Gate** — the required checks before merge: typecheck, lint, tests, build,
45
+ browser test.
46
+
47
+ **Blind review** — reviewers receive the spec and diff, not the implementer's
48
+ self-report.
49
+
50
+ **SCR** — Spec-Compliance Reviewer.
51
+
52
+ **CR** — Code Reviewer.
53
+
54
+ **DAR** — Documentation & Agent-Readability Reviewer.
55
+
56
+ **SEC** — Security Reviewer, triggered for input parsing, attachments, HTML
57
+ injection and file output.
58
+
59
+ **ISTQB edge cases** — test design using equivalence partitioning, boundary
60
+ values, decision tables and state transitions.
package/LICENSE ADDED
@@ -0,0 +1,21 @@
1
+ MIT License
2
+
3
+ Copyright (c) 2026 Tolkira contributors
4
+
5
+ Permission is hereby granted, free of charge, to any person obtaining a copy
6
+ of this software and associated documentation files (the "Software"), to deal
7
+ in the Software without restriction, including without limitation the rights
8
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9
+ copies of the Software, and to permit persons to whom the Software is
10
+ furnished to do so, subject to the following conditions:
11
+
12
+ The above copyright notice and this permission notice shall be included in all
13
+ copies or substantial portions of the Software.
14
+
15
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21
+ SOFTWARE.
package/README.md CHANGED
@@ -1,3 +1,234 @@
1
- # Temporary Holding Version
1
+ # Tolkira
2
2
 
3
- This version is a temporary placeholder for this package. An operational version to replace this has been submitted for review and is awaiting a staged release.
3
+ **One self-contained HTML report for Cucumber JSON, Messages and JUnit XML.**
4
+
5
+ Tolkira turns test results into a single HTML file you can open offline, attach
6
+ to a ticket, or keep as a CI artifact. Screenshots, logs and other attachments
7
+ stay inside the file. There is no server, no CDN and no telemetry.
8
+
9
+ Current release: **v0.1.1** · [npm](https://www.npmjs.com/package/tolkira) · [Changelog](https://github.com/Silasoa/tolkira/blob/main/docs/CHANGELOG.md)
10
+
11
+ > **How this is built:** Tolkira is human-directed and largely AI-generated.
12
+ > The maintainer sets the product direction, reviews changes, runs the release
13
+ > gate and makes release decisions. Most code, tests and documentation are
14
+ > drafted with large language models, then reviewed and tested by humans.
15
+ > AI output can be wrong; please check the report against your own results and
16
+ > open an issue if something looks off. [Read more](#how-this-project-is-built).
17
+
18
+ ## Install
19
+
20
+ ```shell
21
+ npm install --global tolkira
22
+ tolkira --help
23
+ ```
24
+
25
+ Or run it without a global install:
26
+
27
+ ```shell
28
+ npx tolkira --help
29
+ ```
30
+
31
+ On Windows, `npx` can fail when the current folder contains parentheses in its
32
+ path. Use the global install or a folder without parentheses if that happens.
33
+
34
+ ## Quick start
35
+
36
+ ```shell
37
+ # One file, one report
38
+ tolkira test-results/ --out report.html --title "Checkout acceptance"
39
+
40
+ # Add your own metadata and brand colour
41
+ tolkira test-results/ --out report.html \
42
+ --title "Checkout acceptance" \
43
+ --meta "App version=0.3.2" \
44
+ --meta "Environment=Staging" \
45
+ --accent "#1b5e20"
46
+
47
+ # Embed a logo (PNG, JPEG, GIF, WebP or SVG)
48
+ tolkira test-results/ --out report.html --logo path/to/logo.png
49
+ ```
50
+
51
+ Tolkira accepts one or more files and directories. A directory is scanned
52
+ recursively for supported result files. Everything is merged into one report.
53
+
54
+ For CI tools and AI agents, ask for the stable JSON summary instead of scraping
55
+ the HTML:
56
+
57
+ ```shell
58
+ tolkira test-results/ --out report.html --json
59
+ tolkira test-results/ --json
60
+ ```
61
+
62
+ ## What you get
63
+
64
+ - **One file that works offline.** CSS, JavaScript and attachments are inline.
65
+ No network requests, no remote fonts, no analytics.
66
+ - **The failure first.** Failed scenarios are open by default, with the error
67
+ and screenshot where you need them.
68
+ - **Screenshots and attachments that survive.** Images, text, video, audio and
69
+ files are preserved through the legacy JSON adapter.
70
+ - **Search and filters.** Search scenarios, features, steps and tags; filter by
71
+ status and tag; open matching features automatically and highlight matches.
72
+ - **Copy failure.** Copy a failed scenario's feature, failed step and error as
73
+ plain text for a ticket or an agent.
74
+ - **A Details view when you need it.** Status distribution, a feature table,
75
+ run metadata and the Cucumber environment (producer, runtime, OS, CPU,
76
+ protocol).
77
+ - **A stable summary for agents.** `--json` and `summarizeReport()` return a
78
+ versioned schema with counts, durations, scenarios, steps and failure
79
+ messages.
80
+ - **Accessible by default.** Keyboard navigation, visible focus, text status
81
+ labels as well as colour, dark mode and 200% zoom reflow. Accessibility is an
82
+ ongoing effort; feedback is welcome.
83
+ - **Your branding.** Title, metadata, logo and accent colour without forking the
84
+ shell.
85
+
86
+ ## Supported inputs
87
+
88
+ | Input | What works |
89
+ | --- | --- |
90
+ | **Legacy Cucumber JSON** | cucumber-js 2.3.1–13.3.0, Behave 1.3.3, cucumber-jvm 7.22.1 and cucumber-ruby 12.0.1 are tested with real output. Other producers often work; the adapter follows the Cucumber legacy dialect. |
91
+ | **Cucumber Messages (NDJSON)** | cucumber-js 11.2, cucumber-jvm 7.22.1 and cucumber-ruby 12.0.1 fixtures. This is the modern, implementation-independent format. |
92
+ | **JUnit XML** | Basic mode: suites, test cases, statuses, durations, failures, skipped cases and system output. JUnit has no Gherkin steps or screenshots; each test case is shown as one synthetic step. |
93
+
94
+ Multiple files and mixed formats can be merged into one run. See
95
+ [`docs/CUCUMBER-COMPATIBILITY.md`](https://github.com/Silasoa/tolkira/blob/main/docs/CUCUMBER-COMPATIBILITY.md) for the
96
+ tested producer matrix and
97
+ [`docs/JUNIT-XML.md`](https://github.com/Silasoa/tolkira/blob/main/docs/JUNIT-XML.md) for the JUnit limits.
98
+
99
+ ## Why Tolkira exists
100
+
101
+ [`cucumber-html-reporter`](https://github.com/protractor-cucumber-framework/cucumber-html-reporter)
102
+ has been effectively unmaintained since October 2024. It still has a large
103
+ user base, but it loads Bootstrap, jQuery and Google Charts from CDNs, only
104
+ reads legacy JSON, and carries old runtime dependencies. Reports that should
105
+ work offline lose their layout and charts.
106
+
107
+ Tolkira keeps the old input working and fixes the parts that hurt:
108
+
109
+ - one offline file instead of CDN dependencies;
110
+ - legacy JSON **and** modern Messages;
111
+ - JUnit XML as cheap breadth for non-Cucumber teams;
112
+ - screenshots and attachments that stay in the report;
113
+ - a stable JSON summary for agents and CI;
114
+ - a small, reviewed dependency surface.
115
+
116
+ The visual comparison uses the report setup from
117
+ [Seed-Test](https://github.com/adessoSE/Seed-Test). If you are curious about
118
+ Seed-Test, you are welcome to take a look; the comparison notes and screenshots
119
+ are in [`docs/COMPARISON.md`](https://github.com/Silasoa/tolkira/blob/main/docs/COMPARISON.md).
120
+
121
+ ## Migrating from cucumber-html-reporter
122
+
123
+ The old options object is accepted as-is; only the import changes.
124
+
125
+ ```js
126
+ // before
127
+ const reporter = require('cucumber-html-reporter');
128
+
129
+ // after
130
+ import { generate } from 'tolkira/compat/cucumber-html-reporter';
131
+
132
+ await generate({
133
+ jsonFile: 'reports/cucumber.json',
134
+ output: 'reports/report.html',
135
+ brandTitle: 'Checkout acceptance',
136
+ metadata: { Browser: 'Chrome', Environment: 'Staging' },
137
+ });
138
+ ```
139
+
140
+ The shim is experimental. Themes, charts, external screenshots and browser
141
+ launch are not reproduced; they warn and use Tolkira's defaults. It consumes
142
+ the same input JSON, never the old rendered HTML. See
143
+ [`docs/COMPATIBILITY.md`](https://github.com/Silasoa/tolkira/blob/main/docs/COMPATIBILITY.md) for the full mapping and
144
+ caveats.
145
+
146
+ ## Privacy and limits
147
+
148
+ - No network requests, no telemetry, no remote fonts and no CDN.
149
+ - A restrictive Content Security Policy is embedded in every report.
150
+ - Test names, metadata and attachments are treated as untrusted input.
151
+ - Default limits: 100 MB total input, 50 MB per attachment and 100 MB total
152
+ attachments. A warning appears when a report grows beyond 10 MB.
153
+ - JUnit XML is intentionally basic; it is not a replacement for a native
154
+ Playwright, Jest or Allure report.
155
+ - One report is one file. External attachments for very large media are a
156
+ future option.
157
+
158
+ ## Documentation
159
+
160
+ - [`docs/AGENT-INTERFACE.md`](https://github.com/Silasoa/tolkira/blob/main/docs/AGENT-INTERFACE.md) — JSON schema, CLI
161
+ contract and exit codes.
162
+ - [`docs/CUCUMBER-COMPATIBILITY.md`](https://github.com/Silasoa/tolkira/blob/main/docs/CUCUMBER-COMPATIBILITY.md) — tested
163
+ Cucumber producers and versions.
164
+ - [`docs/JUNIT-XML.md`](https://github.com/Silasoa/tolkira/blob/main/docs/JUNIT-XML.md) — JUnit basic-mode mapping and
165
+ limits.
166
+ - [`docs/COMPATIBILITY.md`](https://github.com/Silasoa/tolkira/blob/main/docs/COMPATIBILITY.md) — migration from
167
+ cucumber-html-reporter.
168
+ - [`docs/COMPARISON.md`](https://github.com/Silasoa/tolkira/blob/main/docs/COMPARISON.md) — Seed-Test comparison and
169
+ remaining parity gaps.
170
+
171
+ ## Development
172
+
173
+ Tolkira requires Node.js 22 or newer.
174
+
175
+ ```shell
176
+ npm install
177
+ npx playwright install chromium firefox webkit
178
+ npm run check
179
+ npm run test:e2e
180
+ npm run stress -- 1000
181
+ ```
182
+
183
+ `npm run check` runs typecheck, lint, unit tests, build and the CSS generation
184
+ step. `npm run test:e2e` runs the browser tests in Chromium, Firefox and
185
+ WebKit. `npm run stress -- 1000` builds a 1,000-scenario report and reports size
186
+ and timing.
187
+
188
+ Build once and run the CLI from `dist`:
189
+
190
+ ```shell
191
+ npm run build
192
+ node dist/cli.js test/fixtures/legacy/happy.json \
193
+ --out report.html \
194
+ --title "My report" \
195
+ --meta Browser=Chrome
196
+ ```
197
+
198
+ ## Contributing and feedback
199
+
200
+ Bug reports, accessibility feedback and compatibility fixtures are welcome.
201
+ If a producer or report shape does not work, a small sanitized fixture is the
202
+ most useful thing you can attach. Please do not include secrets or private test
203
+ data.
204
+
205
+ If you want to work on the code, start with [`AGENTS.md`](https://github.com/Silasoa/tolkira/blob/main/AGENTS.md), the
206
+ [`openspec/`](https://github.com/Silasoa/tolkira/tree/main/openspec) change contracts and
207
+ [`docs/agents/review.md`](https://github.com/Silasoa/tolkira/blob/main/docs/agents/review.md). Product and architecture
208
+ notes live in [`docs/PRODUCT.md`](https://github.com/Silasoa/tolkira/blob/main/docs/PRODUCT.md) and
209
+ [`docs/ARCHITECTURE.md`](https://github.com/Silasoa/tolkira/blob/main/docs/ARCHITECTURE.md).
210
+
211
+ ## How this project is built
212
+
213
+ Tolkira is an experiment in AI-assisted open source. The maintainer defines the
214
+ problem, the scope and the design direction; large language models draft most of
215
+ the code, tests and documentation; the maintainer reviews, runs the test and
216
+ release gates, and makes the final release decision.
217
+
218
+ The project does not name specific models in its files on purpose. The
219
+ maintainer configures the models and tooling. This keeps the project
220
+ model-agnostic and avoids turning a temporary tool choice into a permanent
221
+ dependency.
222
+
223
+ AI-generated code can contain mistakes. The test suite and the review process
224
+ reduce that risk, but they do not remove it. If you rely on Tolkira in a
225
+ production pipeline, validate the report against your own results and pin the
226
+ version you use.
227
+
228
+ ## License
229
+
230
+ MIT. See [`LICENSE`](https://github.com/Silasoa/tolkira/blob/main/LICENSE).
231
+
232
+ Tolkira is not affiliated with Cucumber, SmartBear, Seed-Test or any
233
+ test-framework vendor. Cucumber and Gherkin are used descriptively; the project
234
+ uses the MIT-licensed Cucumber Messages format and package.
package/SECURITY.md ADDED
@@ -0,0 +1,30 @@
1
+ # Security Policy
2
+
3
+ ## Supported versions
4
+
5
+ Tolkira is pre-1.0. Security fixes are applied to the latest released version.
6
+
7
+ ## Reporting a vulnerability
8
+
9
+ Please do not open a public issue for a security problem. Use GitHub’s private
10
+ vulnerability reporting / Security Advisories for this repository, or contact
11
+ the maintainers privately.
12
+
13
+ Include:
14
+
15
+ - affected version or commit;
16
+ - a minimal reproduction;
17
+ - impact and suggested fix, if known.
18
+
19
+ ## Scope
20
+
21
+ Tolkira processes untrusted test output. Security-relevant areas include:
22
+
23
+ - HTML/JS injection through test names, metadata, titles and attachment names;
24
+ - active HTML/SVG attachments;
25
+ - path handling for output and attachments;
26
+ - denial of service through very large reports or attachments;
27
+ - dependency vulnerabilities.
28
+
29
+ Tolkira does not send telemetry and does not make network requests. A report
30
+ should remain fully offline.
package/dist/cli.d.ts ADDED
@@ -0,0 +1,5 @@
1
+ export interface CliIO {
2
+ stdout: (text: string) => void;
3
+ stderr: (text: string) => void;
4
+ }
5
+ export declare function runCli(argv: string[], io?: CliIO): Promise<number>;