From c88b2bbb9979a5e6e389c93c172540bdee56f028 Mon Sep 17 00:00:00 2001 From: 0xallam Date: Mon, 25 May 2026 21:28:36 -0700 Subject: [PATCH] Stabilize agent-browser launch and screenshot routing MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit AGENT_BROWSER_ARGS parser splits on commas, so any flag value containing one (--disable-features=A,B, --window-size=1920,1080, --lang=en-US,en) shredded into garbage positionals and Chromium rejected the launch with "Multiple targets are not supported in headless mode". Reduce to a comma-separated list of comma-free flags that keeps the AutomationControlled anti-detection bit. Default screenshot path now resolves inside the workspace root so view_image accepts it; entrypoint pre-creates the dir at runtime (the build-time mkdir is shadowed by the /workspace mount). Skill examples updated to favor the no-arg form, plus brief fallback guidance when view_image is unavailable on text-only models and a viewport-resize note for sites that gate on real desktop dims. Also drop the stale STRIX_DISABLE_BROWSER doc entry — no code reference exists. --- containers/Dockerfile | 4 ++-- containers/docker-entrypoint.sh | 2 ++ docs/advanced/configuration.mdx | 4 ---- strix/skills/tooling/agent_browser.md | 33 ++++++++++++++++++++------- 4 files changed, 29 insertions(+), 14 deletions(-) diff --git a/containers/Dockerfile b/containers/Dockerfile index 6249528..51d5753 100644 --- a/containers/Dockerfile +++ b/containers/Dockerfile @@ -88,10 +88,10 @@ RUN npm install -g retire@latest && \ npm install -g tree-sitter-cli@latest && \ npm install -g agent-browser@0.26.0 -USER pentester ENV AGENT_BROWSER_EXECUTABLE_PATH=/usr/bin/chromium ENV AGENT_BROWSER_USER_AGENT="Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36" -ENV AGENT_BROWSER_ARGS="--disable-blink-features=AutomationControlled --exclude-switches=enable-automation --disable-features=IsolateOrigins,site-per-process,Translate,BlinkGenPropertyTrees --no-first-run --no-default-browser-check --window-size=1920,1080 --lang=en-US,en --disable-infobars --disable-notifications --disable-save-password-bubble --disable-session-crashed-bubble" +ENV AGENT_BROWSER_ARGS="--disable-blink-features=AutomationControlled,--no-first-run,--no-default-browser-check,--lang=en-US" +ENV AGENT_BROWSER_SCREENSHOT_DIR=/workspace/.agent-browser-screenshots RUN /home/pentester/.npm-global/bin/agent-browser doctor --offline --quick RUN set -eux; \ diff --git a/containers/docker-entrypoint.sh b/containers/docker-entrypoint.sh index 6824e3b..f8e179b 100644 --- a/containers/docker-entrypoint.sh +++ b/containers/docker-entrypoint.sh @@ -90,6 +90,8 @@ sudo -u pentester certutil -N -d sql:/home/pentester/.pki/nssdb --empty-password sudo -u pentester certutil -A -n "Testing Root CA" -t "C,," -i /app/certs/ca.crt -d sql:/home/pentester/.pki/nssdb echo "✅ CA added to browser trust store" +mkdir -p /workspace/.agent-browser-screenshots + echo "✅ Container ready" cd /workspace diff --git a/docs/advanced/configuration.mdx b/docs/advanced/configuration.mdx index 4d51f3c..a40b9fc 100644 --- a/docs/advanced/configuration.mdx +++ b/docs/advanced/configuration.mdx @@ -41,10 +41,6 @@ Configure Strix using environment variables or a config file. API key for Perplexity AI. Enables real-time web search during scans for OSINT and vulnerability research. - - Disable browser automation tools. - - Global telemetry default toggle. Set to `0`, `false`, `no`, or `off` to disable both PostHog and OTEL unless overridden by per-channel flags below. diff --git a/strix/skills/tooling/agent_browser.md b/strix/skills/tooling/agent_browser.md index 4682ade..9ef810d 100644 --- a/strix/skills/tooling/agent_browser.md +++ b/strix/skills/tooling/agent_browser.md @@ -18,6 +18,10 @@ wired via ``http_proxy`` / ``https_proxy`` env vars — **do not pass captures all page traffic. Localhost (CDP) traffic is excluded via ``NO_PROXY=localhost,127.0.0.1``. +Default viewport is 1280×720. For sites that gate behavior on real +desktop dimensions (responsive breakpoints, bot fingerprinting), run +``agent-browser viewport 1920 1080`` once per session. + ## The core loop ```bash @@ -37,7 +41,7 @@ next ref interaction. ```bash # Take a screenshot of a page agent-browser open https://example.com -agent-browser screenshot home.png +agent-browser screenshot agent-browser close # Search, click a result, and capture it @@ -48,7 +52,7 @@ agent-browser press Enter agent-browser wait --load networkidle agent-browser snapshot -i # refs now reflect results agent-browser click @e5 # click a result -agent-browser screenshot result.png +agent-browser screenshot ``` The browser stays running across commands so these feel like a single @@ -241,15 +245,22 @@ shell command alone does **not** put the image into your context — chain it with the SDK ``view_image`` tool to actually see it: ```bash -exec_command: agent-browser screenshot /workspace/page.png -view_image: {"path": "/workspace/page.png"} +exec_command: agent-browser screenshot +view_image: {"path": ""} ``` +Default output directory is ``/workspace/.agent-browser-screenshots/``, +which ``view_image`` can read. Prefer the no-arg form (the CLI prints +the full path on stdout — pass that to ``view_image``). If you need a +specific filename, keep it inside that directory or a sibling hidden +dir under ``/workspace``. Never write screenshots to ``/tmp`` — +``view_image`` rejects anything outside the workspace root. + ```bash -agent-browser screenshot # temp path, printed on stdout -agent-browser screenshot page.png # specific path (relative to cwd) -agent-browser screenshot --full full.png # full scroll height -agent-browser screenshot --annotate map.png # numbered labels + legend keyed to snapshot refs +agent-browser screenshot # path printed on stdout +agent-browser screenshot /workspace/.agent-browser-screenshots/page.png +agent-browser screenshot --full # full scroll height +agent-browser screenshot --annotate # numbered labels + legend keyed to snapshot refs ``` `--annotate` is designed for multimodal models: each label `[N]` maps @@ -262,6 +273,12 @@ tokens; screenshots cost more. Use `snapshot` first; reach for layout questions, captchas, custom widgets where the accessibility tree is incomplete). +If ``view_image`` errors back at you (rejected image, "vision not +supported", or similar), you are running on a text-only model — stop +calling it and stop taking screenshots. Drive the page entirely from +`snapshot -i` refs, `eval` for any DOM/JS state you need to read, and +`text @ref` / `get text` for content extraction. + ### Handle multiple pages via tabs ```bash