Skip to content

Commit c7ed16d

Browse files
gmickelgmickel
committed
build: upgrade node-llama-cpp to autoAttempt
* feat/node-llama-cpp-upgrade: test: cover autoAttempt llama initialization build: upgrade node-llama-cpp to autoAttempt
2 parents ce586db + 5f504d1 commit c7ed16d

File tree

14 files changed

+568
-151
lines changed

14 files changed

+568
-151
lines changed

‎.agents/skills/dogfood/SKILL.md‎

Lines changed: 216 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,216 @@
1+
---
2+
name: dogfood
3+
description: Systematically explore and test a web application to find bugs, UX issues, and other problems. Use when asked to "dogfood", "QA", "exploratory test", "find issues", "bug hunt", "test this app/site/platform", or review the quality of a web application. Produces a structured report with full reproduction evidence -- step-by-step screenshots, repro videos, and detailed repro steps for every issue -- so findings can be handed directly to the responsible teams.
4+
allowed-tools: Bash(agent-browser:*), Bash(npx agent-browser:*)
5+
---
6+
7+
# Dogfood
8+
9+
Systematically explore a web application, find issues, and produce a report with full reproduction evidence for every finding.
10+
11+
## Setup
12+
13+
Only the **Target URL** is required. Everything else has sensible defaults -- use them unless the user explicitly provides an override.
14+
15+
| Parameter | Default | Example override |
16+
| -------------------- | ----------------------------------------------------- | ------------------------------------- |
17+
| **Target URL** | _(required)_ | `vercel.com`, `http://localhost:3000` |
18+
| **Session name** | Slugified domain (e.g., `vercel.com` -> `vercel-com`) | `--session my-session` |
19+
| **Output directory** | `./dogfood-output/` | `Output directory: /tmp/qa` |
20+
| **Scope** | Full app | `Focus on the billing page` |
21+
| **Authentication** | None | `Sign in to user@example.com` |
22+
23+
If the user says something like "dogfood vercel.com", start immediately with defaults. Do not ask clarifying questions unless authentication is mentioned but credentials are missing.
24+
25+
Always use `agent-browser` directly -- never `npx agent-browser`. The direct binary uses the fast Rust client. `npx` routes through Node.js and is significantly slower.
26+
27+
## Workflow
28+
29+
```
30+
1. Initialize Set up session, output dirs, report file
31+
2. Authenticate Sign in if needed, save state
32+
3. Orient Navigate to starting point, take initial snapshot
33+
4. Explore Systematically visit pages and test features
34+
5. Document Screenshot + record each issue as found
35+
6. Wrap up Update summary counts, close session
36+
```
37+
38+
### 1. Initialize
39+
40+
```bash
41+
mkdir -p {OUTPUT_DIR}/screenshots {OUTPUT_DIR}/videos
42+
```
43+
44+
Copy the report template into the output directory and fill in the header fields:
45+
46+
```bash
47+
cp {SKILL_DIR}/templates/dogfood-report-template.md {OUTPUT_DIR}/report.md
48+
```
49+
50+
Start a named session:
51+
52+
```bash
53+
agent-browser --session {SESSION} open {TARGET_URL}
54+
agent-browser --session {SESSION} wait --load networkidle
55+
```
56+
57+
### 2. Authenticate
58+
59+
If the app requires login:
60+
61+
```bash
62+
agent-browser --session {SESSION} snapshot -i
63+
# Identify login form refs, fill credentials
64+
agent-browser --session {SESSION} fill @e1 "{EMAIL}"
65+
agent-browser --session {SESSION} fill @e2 "{PASSWORD}"
66+
agent-browser --session {SESSION} click @e3
67+
agent-browser --session {SESSION} wait --load networkidle
68+
```
69+
70+
For OTP/email codes: ask the user, wait for their response, then enter the code.
71+
72+
After successful login, save state for potential reuse:
73+
74+
```bash
75+
agent-browser --session {SESSION} state save {OUTPUT_DIR}/auth-state.json
76+
```
77+
78+
### 3. Orient
79+
80+
Take an initial annotated screenshot and snapshot to understand the app structure:
81+
82+
```bash
83+
agent-browser --session {SESSION} screenshot --annotate {OUTPUT_DIR}/screenshots/initial.png
84+
agent-browser --session {SESSION} snapshot -i
85+
```
86+
87+
Identify the main navigation elements and map out the sections to visit.
88+
89+
### 4. Explore
90+
91+
Read [references/issue-taxonomy.md](references/issue-taxonomy.md) for the full list of what to look for and the exploration checklist.
92+
93+
**Strategy -- work through the app systematically:**
94+
95+
- Start from the main navigation. Visit each top-level section.
96+
- Within each section, test interactive elements: click buttons, fill forms, open dropdowns/modals.
97+
- Check edge cases: empty states, error handling, boundary inputs.
98+
- Try realistic end-to-end workflows (create, edit, delete flows).
99+
- Check the browser console for errors periodically.
100+
101+
**At each page:**
102+
103+
```bash
104+
agent-browser --session {SESSION} snapshot -i
105+
agent-browser --session {SESSION} screenshot --annotate {OUTPUT_DIR}/screenshots/{page-name}.png
106+
agent-browser --session {SESSION} errors
107+
agent-browser --session {SESSION} console
108+
```
109+
110+
Use your judgment on how deep to go. Spend more time on core features and less on peripheral pages. If you find a cluster of issues in one area, investigate deeper.
111+
112+
### 5. Document Issues (Repro-First)
113+
114+
Steps 4 and 5 happen together -- explore and document in a single pass. When you find an issue, stop exploring and document it immediately before moving on. Do not explore the whole app first and document later.
115+
116+
Every issue must be reproducible. When you find something wrong, do not just note it -- prove it with evidence. The goal is that someone reading the report can see exactly what happened and replay it.
117+
118+
**Choose the right level of evidence for the issue:**
119+
120+
#### Interactive / behavioral issues (functional, ux, console errors on action)
121+
122+
These require user interaction to reproduce -- use full repro with video and step-by-step screenshots:
123+
124+
1. **Start a repro video** _before_ reproducing:
125+
126+
```bash
127+
agent-browser --session {SESSION} record start {OUTPUT_DIR}/videos/issue-{NNN}-repro.webm
128+
```
129+
130+
2. **Walk through the steps at human pace.** Pause 1-2 seconds between actions so the video is watchable. Take a screenshot at each step:
131+
132+
```bash
133+
agent-browser --session {SESSION} screenshot {OUTPUT_DIR}/screenshots/issue-{NNN}-step-1.png
134+
sleep 1
135+
# Perform action (click, fill, etc.)
136+
sleep 1
137+
agent-browser --session {SESSION} screenshot {OUTPUT_DIR}/screenshots/issue-{NNN}-step-2.png
138+
sleep 1
139+
# ...continue until the issue manifests
140+
```
141+
142+
3. **Capture the broken state.** Pause so the viewer can see it, then take an annotated screenshot:
143+
144+
```bash
145+
sleep 2
146+
agent-browser --session {SESSION} screenshot --annotate {OUTPUT_DIR}/screenshots/issue-{NNN}-result.png
147+
```
148+
149+
4. **Stop the video:**
150+
151+
```bash
152+
agent-browser --session {SESSION} record stop
153+
```
154+
155+
5. Write numbered repro steps in the report, each referencing its screenshot.
156+
157+
#### Static / visible-on-load issues (typos, placeholder text, clipped text, misalignment, console errors on load)
158+
159+
These are visible without interaction -- a single annotated screenshot is sufficient. No video, no multi-step repro:
160+
161+
```bash
162+
agent-browser --session {SESSION} screenshot --annotate {OUTPUT_DIR}/screenshots/issue-{NNN}.png
163+
```
164+
165+
Write a brief description and reference the screenshot in the report. Set **Repro Video** to `N/A`.
166+
167+
---
168+
169+
**For all issues:**
170+
171+
1. **Append to the report immediately.** Do not batch issues for later. Write each one as you find it so nothing is lost if the session is interrupted.
172+
173+
2. **Increment the issue counter** (ISSUE-001, ISSUE-002, ...).
174+
175+
### 6. Wrap Up
176+
177+
Aim to find **5-10 well-documented issues**, then wrap up. Depth of evidence matters more than total count -- 5 issues with full repro beats 20 with vague descriptions.
178+
179+
After exploring:
180+
181+
1. Re-read the report and update the summary severity counts so they match the actual issues. Every `### ISSUE-` block must be reflected in the totals.
182+
2. Close the session:
183+
184+
```bash
185+
agent-browser --session {SESSION} close
186+
```
187+
188+
3. Tell the user the report is ready and summarize findings: total issues, breakdown by severity, and the most critical items.
189+
190+
## Guidance
191+
192+
- **Repro is everything.** Every issue needs proof -- but match the evidence to the issue. Interactive bugs need video and step-by-step screenshots. Static bugs (typos, placeholder text, visual glitches visible on load) only need a single annotated screenshot.
193+
- **Don't record video for static issues.** A typo or clipped text doesn't benefit from a video. Save video for issues that involve user interaction, timing, or state changes.
194+
- **For interactive issues, screenshot each step.** Capture the before, the action, and the after -- so someone can see the full sequence.
195+
- **Write repro steps that map to screenshots.** Each numbered step in the report should reference its corresponding screenshot. A reader should be able to follow the steps visually without touching a browser.
196+
- **Be thorough but use judgment.** You are not following a test script -- you are exploring like a real user would. If something feels off, investigate.
197+
- **Write findings incrementally.** Append each issue to the report as you discover it. If the session is interrupted, findings are preserved. Never batch all issues for the end.
198+
- **Never delete output files.** Do not `rm` screenshots, videos, or the report mid-session. Do not close the session and restart. Work forward, not backward.
199+
- **Never read the target app's source code.** You are testing as a user, not auditing code. Do not read HTML, JS, or config files of the app under test. All findings must come from what you observe in the browser.
200+
- **Check the console.** Many issues are invisible in the UI but show up as JS errors or failed requests.
201+
- **Test like a user, not a robot.** Try common workflows end-to-end. Click things a real user would click. Enter realistic data.
202+
- **Type like a human.** When filling form fields during video recording, use `type` instead of `fill` -- it types character-by-character. Use `fill` only outside of video recording when speed matters.
203+
- **Pace repro videos for humans.** Add `sleep 1` between actions and `sleep 2` before the final result screenshot. Videos should be watchable at 1x speed -- a human reviewing the report needs to see what happened, not a blur of instant state changes.
204+
- **Be efficient with commands.** Batch multiple `agent-browser` commands in a single shell call when they are independent (e.g., `agent-browser ... screenshot ... && agent-browser ... console`). Use `agent-browser --session {SESSION} scroll down 300` for scrolling -- do not use `key` or `evaluate` to scroll.
205+
206+
## References
207+
208+
| Reference | When to Read |
209+
| ------------------------------------------------------------ | -------------------------------------------------------------------------------------- |
210+
| [references/issue-taxonomy.md](references/issue-taxonomy.md) | Start of session -- calibrate what to look for, severity levels, exploration checklist |
211+
212+
## Templates
213+
214+
| Template | Purpose |
215+
| ---------------------------------------------------------------------------- | --------------------------------------------- |
216+
| [templates/dogfood-report-template.md](templates/dogfood-report-template.md) | Copy into output directory as the report file |

‎.agents/skills/dogfood/references/issue-taxonomy.md‎

Lines changed: 109 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,109 @@
1+
# Issue Taxonomy
2+
3+
Reference for categorizing issues found during dogfooding. Read this at the start of a dogfood session to calibrate what to look for.
4+
5+
## Contents
6+
7+
- [Severity Levels](#severity-levels)
8+
- [Categories](#categories)
9+
- [Exploration Checklist](#exploration-checklist)
10+
11+
## Severity Levels
12+
13+
| Severity | Definition |
14+
| ------------ | ------------------------------------------------------------- |
15+
| **critical** | Blocks a core workflow, causes data loss, or crashes the app |
16+
| **high** | Major feature broken or unusable, no workaround |
17+
| **medium** | Feature works but with noticeable problems, workaround exists |
18+
| **low** | Minor cosmetic or polish issue |
19+
20+
## Categories
21+
22+
### Visual / UI
23+
24+
- Layout broken or misaligned elements
25+
- Overlapping or clipped text
26+
- Inconsistent spacing, padding, or margins
27+
- Missing or broken icons/images
28+
- Dark mode / light mode rendering issues
29+
- Responsive layout problems (viewport sizes)
30+
- Z-index stacking issues (elements hidden behind others)
31+
- Font rendering issues (wrong font, size, weight)
32+
- Color contrast problems
33+
- Animation glitches or jank
34+
35+
### Functional
36+
37+
- Broken links (404, wrong destination)
38+
- Buttons or controls that do nothing on click
39+
- Form validation that rejects valid input or accepts invalid input
40+
- Incorrect redirects
41+
- Features that fail silently
42+
- State not persisted when expected (lost on refresh, navigation)
43+
- Race conditions (double-submit, stale data)
44+
- Broken search or filtering
45+
- Pagination issues
46+
- File upload/download failures
47+
48+
### UX
49+
50+
- Confusing or unclear navigation
51+
- Missing loading indicators or feedback after actions
52+
- Slow or unresponsive interactions (>300ms perceived delay)
53+
- Unclear error messages
54+
- Missing confirmation for destructive actions
55+
- Dead ends (no way to go back or proceed)
56+
- Inconsistent patterns across similar features
57+
- Missing keyboard shortcuts or focus management
58+
- Unintuitive defaults
59+
- Missing empty states or unhelpful empty states
60+
61+
### Content
62+
63+
- Typos or grammatical errors
64+
- Outdated or incorrect text
65+
- Placeholder or lorem ipsum content left in
66+
- Truncated text without tooltip or expansion
67+
- Missing or wrong labels
68+
- Inconsistent terminology
69+
70+
### Performance
71+
72+
- Slow page loads (>3s)
73+
- Janky scrolling or animations
74+
- Large layout shifts (content jumping)
75+
- Excessive network requests (check via console/network)
76+
- Memory leaks (page slows over time)
77+
- Unoptimized images (large file sizes)
78+
79+
### Console / Errors
80+
81+
- JavaScript exceptions in console
82+
- Failed network requests (4xx, 5xx)
83+
- Deprecation warnings
84+
- CORS errors
85+
- Mixed content warnings
86+
- Unhandled promise rejections
87+
88+
### Accessibility
89+
90+
- Missing alt text on images
91+
- Unlabeled form inputs
92+
- Poor keyboard navigation (can't tab to elements)
93+
- Focus traps
94+
- Insufficient color contrast
95+
- Missing ARIA attributes on dynamic content
96+
- Screen reader incompatible patterns
97+
98+
## Exploration Checklist
99+
100+
Use this as a guide for what to test on each page/feature:
101+
102+
1. **Visual scan** -- Take an annotated screenshot. Look for layout, alignment, and rendering issues.
103+
2. **Interactive elements** -- Click every button, link, and control. Do they work? Is there feedback?
104+
3. **Forms** -- Fill and submit. Test empty submission, invalid input, and edge cases.
105+
4. **Navigation** -- Follow all navigation paths. Check breadcrumbs, back button, deep links.
106+
5. **States** -- Check empty states, loading states, error states, and full/overflow states.
107+
6. **Console** -- Check for JS errors, failed requests, and warnings.
108+
7. **Responsiveness** -- If relevant, test at different viewport sizes.
109+
8. **Auth boundaries** -- Test what happens when not logged in, with different roles if applicable.

‎.agents/skills/dogfood/templates/dogfood-report-template.md‎

Lines changed: 53 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,53 @@
1+
# Dogfood Report: {APP_NAME}
2+
3+
| Field | Value |
4+
| ----------- | -------------- |
5+
| **Date** | {DATE} |
6+
| **App URL** | {URL} |
7+
| **Session** | {SESSION_NAME} |
8+
| **Scope** | {SCOPE} |
9+
10+
## Summary
11+
12+
| Severity | Count |
13+
| --------- | ----- |
14+
| Critical | 0 |
15+
| High | 0 |
16+
| Medium | 0 |
17+
| Low | 0 |
18+
| **Total** | **0** |
19+
20+
## Issues
21+
22+
<!-- Copy this block for each issue found. Interactive issues need video + step-by-step screenshots. Static issues (typos, visual glitches) only need a single screenshot -- set Repro Video to N/A. -->
23+
24+
### ISSUE-001: {Short title}
25+
26+
| Field | Value |
27+
| --------------- | -------------------------------------------------------------------------- |
28+
| **Severity** | critical / high / medium / low |
29+
| **Category** | visual / functional / ux / content / performance / console / accessibility |
30+
| **URL** | {page URL where issue was found} |
31+
| **Repro Video** | {path to video, or N/A for static issues} |
32+
33+
**Description**
34+
35+
{What is wrong, what was expected, and what actually happened.}
36+
37+
**Repro Steps**
38+
39+
<!-- Each step has a screenshot. A reader should be able to follow along visually. -->
40+
41+
1. Navigate to {URL}
42+
![Step 1](screenshots/issue-001-step-1.png)
43+
44+
2. {Action -- e.g., click "Settings" in the sidebar}
45+
![Step 2](screenshots/issue-001-step-2.png)
46+
47+
3. {Action -- e.g., type "test" in the search field and press Enter}
48+
![Step 3](screenshots/issue-001-step-3.png)
49+
50+
4. **Observe:** {what goes wrong -- e.g., the page shows a blank white screen instead of search results}
51+
![Result](screenshots/issue-001-result.png)
52+
53+
---

‎.claude/skills/dogfood‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
../../.agents/skills/dogfood

‎.flow/epics/fn-32-upgrade-node-llama-cpp-to-autoattempt.json‎

Lines changed: 13 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,13 @@
1+
{
2+
"branch_name": "feat/node-llama-cpp-upgrade",
3+
"created_at": "2026-03-08T14:52:44.921727Z",
4+
"depends_on_epics": [],
5+
"id": "fn-32-upgrade-node-llama-cpp-to-autoattempt",
6+
"next_task": 1,
7+
"plan_review_status": "unknown",
8+
"plan_reviewed_at": null,
9+
"spec_path": ".flow/specs/fn-32-upgrade-node-llama-cpp-to-autoattempt.md",
10+
"status": "open",
11+
"title": "Upgrade node-llama-cpp to autoAttempt build path",
12+
"updated_at": "2026-03-08T14:52:56.356149Z"
13+
}

0 commit comments

Comments
 (0)