Dogfood

Dogfood

Systematically explore a web application, find issues, and produce a report with full reproduction evidence for every finding.

Uses agent-browser, a headless browser automation CLI.

Setup

Only the Target URL is required. Everything else has sensible defaults.

| Parameter | Default | Example override | |-----------|---------|-----------------| | Target URL | (required) | vercel.com, http://localhost:3000 | | Session name | Slugified domain (e.g., vercel.com -> vercel-com) | --session my-session | | Output directory | ./dogfood-output/ | Output directory: /tmp/qa | | Scope | Full app | Focus on the billing page | | Authentication | None | Sign in to user@example.com |

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.

Always use agent-browser directly (not npx agent-browser). The direct binary uses the fast Rust client.

Before starting, check prerequisites:

command -v agent-browser || true
command -v surge || true  # only needed if user asks to publish

If agent-browser is unavailable, do not stop. Use browser tools or Playwright as a fallback and read references/playwright-surge-fallback.md. The goal is evidence-backed QA, not loyalty to one browser driver.

Workflow

1. Initialize    Set up session, output dirs, report file
2. Authenticate  Sign in if needed, save state
3. Orient        Navigate to starting point, take initial snapshot
4. Explore       Systematically visit pages and test features
5. Document      Screenshot + record each issue as found
6. Wrap up       Update summary counts, close session

1. Initialize

mkdir -p {OUTPUT_DIR}/screenshots {OUTPUT_DIR}/videos

Copy the report template into the output directory and fill in the header fields:

cp {SKILL_DIR}/templates/dogfood-report-template.md {OUTPUT_DIR}/report.md

Where {SKILL_DIR} is the directory containing this SKILL.md file.

Start a named session:

agent-browser --session {SESSION} open {TARGET_URL}
agent-browser --session {SESSION} wait --load networkidle

2. Authenticate

If the app requires login:

agent-browser --session {SESSION} snapshot -i
# Identify login form refs, fill credentials
agent-browser --session {SESSION} fill @e1 "{EMAIL}"
agent-browser --session {SESSION} fill @e2 "{PASSWORD}"
agent-browser --session {SESSION} click @e3
agent-browser --session {SESSION} wait --load networkidle

For OTP/email codes: ask the user, wait for their response, then enter the code.

After successful login, save state for potential reuse:

agent-browser --session {SESSION} state save {OUTPUT_DIR}/auth-state.json

3. Orient

Take an initial annotated screenshot and snapshot to understand the app structure:

agent-browser --session {SESSION} screenshot --annotate {OUTPUT_DIR}/screenshots/initial.png
agent-browser --session {SESSION} snapshot -i

Identify the main navigation elements and map out the sections to visit.

4. Explore

Read [references/issue-taxonomy.md](references/issue-taxonomy.md) for the full list of what to look for and the exploration checklist.

Strategy: work through the app systematically:

• Start from the main navigation. Visit each top-level section.
• Within each section, test interactive elements: click buttons, fill forms, open dropdowns/modals.
• Check edge cases: empty states, error handling, boundary inputs.
• Try realistic end-to-end workflows (create, edit, delete flows).
• Check the browser console for errors periodically.

At each page:

agent-browser --session {SESSION} snapshot -i
agent-browser --session {SESSION} screenshot --annotate {OUTPUT_DIR}/screenshots/{page-name}.png
agent-browser --session {SESSION} errors
agent-browser --session {SESSION} console

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.

5. Document Issues (Repro-First)

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.

Every issue must be reproducible. When you find something wrong, do not just note it: prove it with evidence.

Choose the right level of evidence for the issue:

Interactive / behavioral issues (functional, ux, console errors on action)

These require user interaction to reproduce. Use full repro with video and step-by-step screenshots:

1. Start a repro video before reproducing:

agent-browser --session {SESSION} record start {OUTPUT_DIR}/videos/issue-{NNN}-repro.webm

1. Walk through the steps at human pace. Pause 1-2 seconds between actions so the video is watchable. Take a screenshot at each step:

agent-browser --session {SESSION} screenshot {OUTPUT_DIR}/screenshots/issue-{NNN}-step-1.png
sleep 1
# Perform action (click, fill, etc.)
sleep 1
agent-browser --session {SESSION} screenshot {OUTPUT_DIR}/screenshots/issue-{NNN}-step-2.png
sleep 1
# ...continue until the issue manifests

1. Capture the broken state. Pause so the viewer can see it, then take an annotated screenshot:

sleep 2
agent-browser --session {SESSION} screenshot --annotate {OUTPUT_DIR}/screenshots/issue-{NNN}-result.png

1. Stop the video:

agent-browser --session {SESSION} record stop

1. Write numbered repro steps in the report, each referencing its screenshot.

Static / visible-on-load issues (typos, placeholder text, clipped text, misalignment, console errors on load)

These are visible without interaction. A single annotated screenshot is sufficient. No video, no multi-step repro:

agent-browser --session {SESSION} screenshot --annotate {OUTPUT_DIR}/screenshots/issue-{NNN}.png

Write a brief description and reference the screenshot in the report. Set Repro Video to N/A.

---

For all issues:

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.

1. Increment the issue counter (ISSUE-001, ISSUE-002, ...).

6. Wrap Up

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.

After exploring:

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.
2. If using agent-browser, close the session:

agent-browser --session {SESSION} close

1. If the user asked to publish, create a readable index.html in the output directory, keep report.md and artifacts alongside it, deploy with Surge, then verify HTTP 200 before replying. See references/playwright-surge-fallback.md for the exact Surge pattern.
2. Tell the user the report is ready and summarize findings: total issues, breakdown by severity, most critical items, and the public URL if published.

Guidance

• 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.
• 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.
• For interactive issues, screenshot each step. Capture the before, the action, and the after so someone can see the full sequence.
• 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.
• 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.
• 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.
• 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.
• 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.
• Check the console. Many issues are invisible in the UI but show up as JS errors or failed requests.
• Test like a user, not a robot. Try common workflows end-to-end. Click things a real user would click. Enter realistic data.
• Type like a human. When filling form fields during video recording, use type instead of fill (types character-by-character). Use fill only outside of video recording when speed matters.
• 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.
• Be efficient with commands. Batch multiple agent-browser commands in a single shell call when they are independent. Use agent-browser --session {SESSION} scroll down 300 for scrolling.
• Also test the backend directly — but verify your payloads. When the app hits an API, curl the endpoints independently. But first, inspect what the frontend actually sends (use eval to read fetch call arguments, or check network requests). A 500 from a malformed test payload is YOUR bug, not the app's. If you can't determine the correct payload shape, note it as "API returned error with test payload; may be payload mismatch" rather than declaring "API is broken." Reporting false positives destroys credibility.
• Distinguish automation artifacts from real bugs. Some issues only appear because of how agent-browser interacts with the page (e.g., $ signs eaten by type, buttons unclickable via ref but fine for real users). Before reporting, ask: "Would a human user see this?" If unsure, mark as "unconfirmed — may be automation artifact" and suggest manual verification.

Pitfalls (agent-browser quirks)

These are real issues encountered during dogfood sessions. Read before starting.

• wait --load networkidle often times out on SPAs. Skip it if it fails and proceed with screenshots/snapshots. The page is usually ready.
• press takes only a key, not a selector. Correct: press Enter. Wrong: press @e12 Enter. To press a key while focused on an element, click or focus the element first, then press.
• There is no js, exec, or viewport command. Use eval to run JavaScript. There is no built-in viewport resize; use eval "window.resizeTo(w, h)" (may not work in headless) or accept the default viewport.
• type may interpret $ as special. The type command sends keystrokes and $ followed by digits can be swallowed or misinterpreted. If testing dollar amounts, verify the displayed text matches what you typed (use eval or get text to read back the field value). Use fill instead of type when exact fidelity matters more than human-like typing. Flag dollar-sign issues as "unconfirmed — may be automation artifact" unless verified in a real browser.
• Ref-based click can fail with "matched N elements". This happens when the page has changed since the last snapshot. Always re-run snapshot -i to get fresh refs. If a button still can't be clicked by ref, use a CSS selector or eval "document.querySelector('...').click()" as fallback.
• Be efficient with commands. Batch multiple agent-browser commands in a single shell call when they are independent. Use agent-browser --session {SESSION} scroll down 300 for scrolling.
• Also test the backend directly — but verify your payloads. When the app hits an API, curl the endpoints independently. But first, inspect what the frontend actually sends (use eval to read fetch call arguments, or check network requests). A 500 from a malformed test payload is YOUR bug, not the app's. If you can't determine the correct payload shape, note it as "API returned error with test payload; may be payload mismatch" rather than declaring "API is broken." Reporting false positives destroys credibility.
• Distinguish automation artifacts from real bugs. Some issues only appear because of how agent-browser interacts with the page (e.g., $ signs eaten by type, buttons unclickable via ref but fine for real users). Before reporting, ask: "Would a human user see this?" If unsure, mark as "unconfirmed — may be automation artifact" and suggest manual verification.

Pitfalls (agent-browser quirks)

These are real issues encountered during dogfood sessions. Read before starting.

• wait --load networkidle often times out on SPAs. Skip it if it fails and proceed with screenshots/snapshots. The page is usually ready.
• press takes only a key, not a selector. Correct: press Enter. Wrong: press @e12 Enter. To press a key while focused on an element, click or focus the element first, then press.
• There is no js, exec, or viewport command. Use eval to run JavaScript. There is no built-in viewport resize; use eval "window.resizeTo(w, h)" (may not work in headless) or accept the default viewport.
• type may interpret $ as special. The type command sends keystrokes and $ followed by digits can be swallowed or misinterpreted. If testing dollar amounts, verify the displayed text matches what you typed (use eval or get text to read back the field value). Use fill instead of type when exact fidelity matters more than human-like typing. Flag dollar-sign issues as "unconfirmed — may be automation artifact" unless verified in a real browser.
• Ref-based click can fail with "matched N elements". This happens when the page has changed since the last snapshot. Always re-run snapshot -i to get fresh refs. If a button still can't be clicked by ref, use a CSS selector or eval "document.querySelector('...').click()" as fallback.
• The submit/send button pattern. Many chat UIs have submit buttons inside textareas that are hard to target. Fallback sequence: (1) try click @ref, (2) try press Enter after focusing the textarea, (3) use eval to find and click the button via DOM query.

References

| Reference | When to Read | |-----------|--------------| | [references/issue-taxonomy.md](references/issue-taxonomy.md) | Start of session: calibrate what to look for, severity levels, exploration checklist | | [references/backend-5xx-triage.md](references/backend-5xx-triage.md) | When an API probe returns a 5xx — classify crash vs. isolated error vs. transient restart (poll /health, time it, reproduce), spot the async-route-calls-sync-ORM crash, and don't trust an open PR to close the bug you reproduced | | [references/design-critique.md](references/design-critique.md) | When the session scope includes UI/UX quality — run after functional exploration to score heuristics, flag visual/color/typography issues, and produce a Critical/Important/Polish triage | | [references/playwright-surge-fallback.md](references/playwright-surge-fallback.md) | When agent-browser is unavailable/failing, when using Playwright/browser tools as the driver, or when publishing a public dogfood report to Surge |

Templates

| Template | Purpose | |----------|---------| | [templates/dogfood-report-template.md](templates/dogfood-report-template.md) | Report structure copied to output dir at start |

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: exiao
• Source: exiao/pm-skills
• License: Apache-2.0

Install and usage instructions live in the source repository linked above.