Now an MCP server connect Claude Code, Cursor, Codex

Pin feedback on the live website.
Use Flag It MCP to fix it on your Claude Code

Flag It captures every detail: screenshot, code context, and console errors, then sends it straight to Claude Code, Cursor, or your developer via Jira and Linear. No screenshots in Slack. No back-and-forth. Just context so good the fix happens first try.

7-day free trial all features no credit card
broken
acme.com/checkout
Checkout
Step 3 of 3 · Payment
Submit
1
Flag · Maya
“Submit” doesn't tell users this goes to payment. Red feels like an error.
claude code · via mcp
claude code · useflagit.com
$ claude “Fix the open flags for useflagit.com”
Connecting to Flag It MCP
Pulled 3 open flags with full context.
checkout button: label + color
nav overflow on mobile
pricing card alignment
// editing components/Checkout.tsx
bg #EF4444 #10B981
No screenshots. No follow-up questions. First try.
1. Flag It captures·2. MCP serves context·3. Claude Code ships
Three ways to fix

Flag it once. Fix it however you work.

Every flag carries the same rich context. From there, the fix happens however your team works: automatically via MCP, through a Jira or Linear ticket, or as a GitHub PR the agent opens for you.

Every flag includes
Captured automatically · zero effort from the person flagging
Annotated screenshot
CSS selector
Element HTML
Computed styles
Console errors
Failed network requests
Browser + device
Framework detection
AI categorization
Effort estimate
Recommended fix
Screen recording
Path 1 · MCP
Pin a bug. Claude Code fixes it.

Flag It is an MCP server. Connect Claude Code, Cursor, or Codex. They pull the flag context directly and make the change. No copy-pasting. No translating screenshots into code.

$ claude “Fix the open flags”
→ connects to Flag It MCP
→ pulls 3 flags · fixes each
JL
Path 2 · Tickets
The best ticket your developer has ever received.

One click creates a Jira or Linear ticket with the full context attached: screenshot, console errors, AI triage, effort estimate, recommended fix. They open it and have everything they need.

Full dev context attached
Or paste it into their own AI tool
Effort + priority pre-set
GPath 3 · Beta
Let the agent open the PR.

For approved fixes, the Flag It agent localizes the change in your repo, writes the patch, and opens a draft GitHub PR with the full context and a blast-radius check. You review and merge.

branchfix/checkout-button
files1 changed · +2 −2
statusdraft PR open
The full loop

From broken to fixed
in three steps.

The feedback-to-fix loop used to take 2 weeks. Now it takes minutes, with the PM in control.
Step 01
Pin the issue on any live page

Anyone on your team pins an issue on the live website. They type what's wrong. Flag It captures the screenshot, the element, console errors, and device details automatically.

acme.com/checkout
Checkout
Step 3 of 3 · Payment
Submit
1
New flag · 2 sec ago
Button is red and says “Submit”. Feels like an error. Should it go to payment?
⌘+Enter to flag
Screenshot ✓
Element ✓
Console ✓
Device ✓
Who it's for

Built for teams that
ship fast.

PMs who build with AI
20 min
→ minutes
The pain

You ship fixes yourself with Claude Code or Cursor. But you waste 20 minutes per bug translating feedback into something the AI can work with.

with Flag It

Flag It gives you structured context your coding tool understands natively, via MCP or copy-paste. Fix in minutes, not hours.

Agency PMs
8 projects
one inbox
The pain

You manage 8 client projects. Feedback comes via email, WhatsApp, and Loom. Scattered everywhere.

with Flag It

Flag It centralizes it. Every flag becomes a Jira or Linear ticket with full developer context, or feeds straight into your coding tool via MCP.

QA teams & developers
50 / wk
one click each
The pain

Your QA team captures 50 issues a week. They screenshot each one, write up repro steps, file tickets.

with Flag It

Flag It does all of that in one click. The developer opens the ticket, or connects their AI coding tool, and has the selector, the HTML, the console errors, and a recommended fix.

Why context changes everything

Claude Code and Cursor are only as good as the context they get.

A screenshot in Slack gives Claude Code nothing to work with. A Flag It flag gives it everything.

01What Flag It captures

Screenshot + pin location. CSS selector. Element HTML and computed styles. Console errors and failed network requests from the last 60 seconds. Framework detection. Browser, device, and viewport.

bugux confusioncopyperformancefix nownice to havequick fixmulti-day
02Why it matters for AI tools

Feed a flag into Claude Code or Cursor and it doesn't ask follow-up questions. It has the selector, the HTML, the styles, and a recommended fix. It knows exactly which file to edit. First try.

// recommended fix
.btn-submit {
background: #EF4444 #10B981;
label: "Submit" "Continue to Payment";
}
03MCP: the automatic connection

Flag It is an MCP server. Claude Code and Cursor connect natively. They ask 'what are the open flags?' and get structured context they can act on. No copy-pasting from Jira. No translating screenshots.

// claude code → flag it mcp
→ list_flags()
returns 3 open flags · full context
→ get_flag(id)
selector · html · styles · console
→ resolve_flag(id)
04The agent opens the PRBeta

For approved fixes, the agent localizes the change, writes the patch, and opens a draft PR in your GitHub repo with full context and a blast-radius check. You review and merge.

branchfix/checkout-button
files1 changed · +2 −2
statusdraft PR open
checksblast-radius · policy ✓
Integrations

Connects to how you already work.

AI coding tools · via MCP
C
Claude Code
Pulls flags via MCP, makes the fix
Cu
Cursor
Native MCP context for AI fixes
Cx
Codex
Structured flag context on demand
W
Windsurf
Any MCP-compatible coding tool
Project management · capture
J
Jira
One-click ticket with full context
L
Linear
One-click ticket with full context
C
Chrome
Browser extension, any website
Beta
G
GitHub
Agent opens draft PRs from fixes
Coming soon
S
Slack
Notifications when flags are created
Every flag carries the same context
Screenshot
CSS selector
Console errors
AI triage
Effort estimate
Browser + device
Framework detection
Recommended fix
Data mode

Not just for
product teams.

Flag It works on BI dashboards too. Switch to Data Mode and your business heads, product leaders, and data leads can flag issues on Metabase, Hex, or any analytics dashboard. Structured feedback: what's wrong, what was expected, which filters. It all goes straight to your data analysts.

MetabaseHexLookerTableauModeAny dashboard
metrics.acme.com · data mode
1
Flag · structured for data teams
What's wrong
Spike on day 6 looks wrong
Expected
Same range as adjacent days
Filters
last_30d · region: NA · paid
Pricing

Simple pricing.
Start free.

7-day free trial on all plans. No credit card required. All features included on every plan. Just pick your usage level.

Starter
$29/mo
Start free trial →
Feedback items / month100
Projects3
Action seats2
Feedback givers20
AI triage + recommended fix
MCP (Claude Code, Cursor)
Jira + Linear
Console snapshot
Screen recording
Priority support·
Most popular
Growth
$79/mo
Start free trial →
Feedback items / month500
ProjectsUnlimited
Action seats5
Feedback giversUnlimited
AI triage + recommended fix
MCP (Claude Code, Cursor)
Jira + Linear
Console snapshot
Screen recording
Priority support·
Scale
$149/mo
Start free trial →
Feedback items / month1,000
ProjectsUnlimited
Action seats10
Feedback giversUnlimited
AI triage + recommended fix
MCP (Claude Code, Cursor)
Jira + Linear
Console snapshot
Screen recording
Priority support
Overage $0.10 / additional item·Extra action seat $10 / mo·Trial 7 days, all features
Built different

“Built by a PM who got tired of waiting for developers.”

Flag It started because the feedback-to-fix loop was the slowest part of shipping software. Now it's the fastest.

Founder · Flag It
PM who codes with AI
2 hrs
QA-to-fix cycle
Down from 2 weeks
71%
PM time spent fixing
Up from 26%
Early users
Agency studiosSolo PMsQA leadsAI buildersData teams
Final word

Stop describing bugs.
Start fixing them.

You've spent enough time screenshotting, annotating, and explaining. Flag It captures the context. Claude Code, Cursor, or your developer handles the rest.

7 days · all features · no credit card