hermes-chrome
Skill0 starsChanged 3 months agoMITView on GitHub
Drive the user's daily Chrome via Hermes Chrome (extension + local bridge :19876 + CLI). Use for real cookies/SSO, workspace tabs, capture, cookie-aware download, local URL/file safety checks, light DOM ops without hijacking the active tab. Prefer over headless when live daily Chrome is required.
In this repository66 other files it gives its agents.
- adb-android-app-opsSkill
- agents-sdkSkill
- api-documentationSkill
- atlasSkill
- backend-api-auth-guardrailsSkill
- browser-useSkill
- bundleSkill
- changelogSkill
- chatgpt-appsSkill
- chrome-devtools-testSkill
- chronicleSkill
- cloudflare-deploySkill
- cloudflare-email-serviceSkill
- cloudflare-one-migrationsSkill
- cloudflare-oneSkill
- cloudflareSkill
- computer-useSkill
- create-planSkill
- database-analysisSkill
- database-data-guardrailsSkill
- debug-memory-leakSkill
- deep-initSkill
- deploySkill
- develop-web-gameSkill
- dev-workflowSkill
- docSkill
- durable-objectsSkill
- explainSkill
- figmaSkill
- firmware-feature-writerSkill
- fix-bugSkill
- fix-lintSkill
- frontend-mobile-uiux-designerSkill
- frontend-ui-guardrailsSkill
- gemini-cliSkill
- grok-x-searchSkill
- image-file-readerSkill
- imagegenSkill
- infra-ops-mobile-guardrailsSkill
- layout-reviewSkill
- localizeSkill
- multimodal-lookerSkill
- openai-docsSkill
- optimizeSkill
- parallel-devSkill
- plan-modeSkill
- playwrightSkill
- project-planningSkill
- qa-testerSkill
- refactorSkill
- review-changesSkill
- review-fix-shipSkill
- rust-programmerSkill
- sandbox-sdkSkill
- screenshotSkill
- security-auditSkill
- security-scanSkill
- security-threat-modelSkill
- source-command-collab-startSkill
- turnstile-spinSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 23
Do not osascript / tell Google Chrome to activate
- Distinctive: line 24
Do not rewrite the user's active tab
- Very specific: line 27
Prefer check-url before download.
- Distinctive: line 28
No cloud threat APIs by default
- Very specific: line 85
If ping version lags repo, tell user: Reload unpacked/CWS + click icon.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Hermes Chrome (Grok skill)
When to use
- Need user's real Chrome cookies / SSO / already-open tabs
- Logged-in capture, SSO smoke, already-open workspace tabs
- Download with cookie jar, local URL/file safety checks
- User says hermes-chrome / daily Chrome / bridge 19876
When NOT to use
- Public pages with no login → Hermes headless
browser_*/ Playwright headless - Own-app E2E / layout → Chrome DevTools MCP / Playwright / browser-e2e
- Isolated headed browser without daily cookies → Agent Chrome
:9333
Hard rules
- Do not
osascript/tell Google Chrome to activate - Do not rewrite the user's active tab
- Fail-fast: if
pingtimes out once, stop the chain; ask user to Reload + click icon - Gate with health: prefer
extension_connected: trueon/v1/health - Prefer
check-urlbefore download; do not--forceblock findings without user ask - No cloud threat APIs by default
Paths
| Piece | Path |
| CLI | ~/.hermes/scripts/hermes-chrome.sh or repo scripts/hermes-chrome.sh |
| Repo | local Hermes Chrome checkout (CLI usually under ~/.hermes/scripts/) |
| Bridge | http://127.0.0.1:19876 |
| Runtime | ~/.hermes/run/hermes-chrome/ |
| Downloads | ~/.hermes/run/hermes-chrome/downloads/ |
Ops loop
CLI=~/.hermes/scripts/hermes-chrome.sh
# 1) Bridge
$CLI bridge-status
# extension_connected should be true; if bridge down: $CLI install-launchd or bridge-start
# 2) Extension gate (fail-fast)
$CLI ping
# expect: "extension":"hermes-chrome","version":"1.4.1"+
# 3) Work
$CLI start 'https://example.com/'
$CLI list-tabs --group
$CLI navigate 'https://example.org/' --tab-id <id>
$CLI capture --prefer active --out /tmp/page.png
$CLI eval --tab-id <id> --expr 'document.title'
$CLI page-assets --tab-id <id>
$CLI check-tab-links --tab-id <id>
$CLI --json ping
$CLI check-url 'https://example.com/report.pdf'
$CLI download 'https://example.com/report.pdf' # direct
$CLI download 'https://app.example/private' --cookies # cookie jar
$CLI analyze ~/.hermes/run/hermes-chrome/downloads/report.pdf
$CLI status
$CLI stop # closes Hermes workspace tabsCapture / download notes
captureVisibleTabmay briefly activate the target tab; put capture targets in a separate window when possible- Optional title filters can target known window titles; do not hardcode product-specific symbols
downloadruns check → save → analyze unless--no-check/--no-analyze/--forceanalyzeis local heuristics (not antivirus)
Version gate
| Version | Capability |
| 1.2.0+ | capture, list-tv |
| 1.3.0+ | list-tabs, navigate, eval, click, type, health extension_last_seen |
| 1.4.1+ | fetch_url (cookies download), page-assets, CLI check-url/download/analyze |
If ping version lags repo, tell user: Reload unpacked/CWS + click icon.
Install / recovery
$CLI install-help
$CLI install-launchd # macOS login + KeepAlive
# chrome://extensions → Load unpacked → <repo>/extension → click iconDoes it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

