MCP server · Use cases

flconsole MCP use cases: 4 ways to find Upwork jobs with your AI agent

flconsole is a free hosted MCP server that gives your AI agent new Upwork jobs by your search, job and client filters. Here are four setups that work today: a freelancer in the Claude app with no terminal, a developer running Claude Code on a loop, an agency sharing jobs with its team in Telegram, and niche research with real job counts.

4 use cases · Prompts to copy · Free, no API key

On this page

Which use case fits you?

Each one uses the same free server and the same sign-in through Telegram. Pick by where you work and what you want back.

flconsole MCP use cases compared
Use caseWhoRuns inYou get
1. No terminalA freelancer who works in chatClaude appA shortlist with reasons and proposal drafts whenever you ask
2. On a loopA developerClaude CodeNew jobs that fit every 15 minutes, a weekly tune-up with numbers
3. Team feedsAn agency or a teamAny agent + TelegramJobs for each specialization in the team chat
4. Niche researchAnyone choosing a nicheAny agentJobs a day per niche over 7 days, then a feed for your agent

Use case 1

Freelancer in the Claude app: a shortlist without a terminal

Mia is a product designer. She works in the Claude app, not in a terminal, and wants Claude to read new Upwork jobs for her and keep only the ones worth a proposal.

Who
Freelancer, no code
Goal
Fewer jobs to read, better proposals
Works in
Claude app, web or desktop
Way
Feeds for your agent + memory
  1. Step 1: Connect flconsole as a custom connector

    In Customize → Connectors, click + → Add custom connector, paste the URL and sign in through Telegram: tap Allow in the bot and type the 4-digit code.

    https://mcp.lite.flconsole.com/mcp
  2. Step 2: Paste the prompt into a chat or a Claude project

    Put it in a project's instructions and every chat in it works this way. Claude saves the search as a feed for itself, with its notes, on flconsole, so the next chat picks up where this one stopped.

    Ask your agent: paste this prompt
    Use flconsole as my Upwork job scout. Call get_agent_context first, every chat.
    No feed yet? Set one up with me in 3 steps: what to search for, job and client filters, what fits me. Count matching jobs at each step and save it on flconsole with create_feed, alerts off.
    When I say "check", find new jobs in my feeds and show only the ones that fit: title, rate or budget, client, one line on why it fits.
    When I pick a job, draft a short proposal: the client's problem, one result of mine, a first step, one question. Never send anything for me.
    Save what you learn about me with save_agent_instructions.

    Prefer it installed? Upload the upwork-job-hunter skill in Customize → Skills and say “Set up my Upwork job search.”

  3. Step 3: Say “check” when you sit down to work

    Your AI agent

    check

    flconsole · get_agent_context1 feed: "saas ux", notes about you

    flconsole · find_jobs (feed)7 new jobs

    2 worth a look:

    1. Figma redesign of a SaaS onboarding flow — $1,500 fixed, US client, $18K spentFits: SaaS onboarding, your main case study
    2. UX audit and new dashboard for a fintech app — $50-70/hr, Germany, payment verifiedFits: dashboards and fintech, both in your notes

    Skipped 5: logo-only work, unpaid test tasks, a full-time role.

    Draft a proposal for the first one

    Draft, not sent: “Your onboarding loses people before they reach the first project. I rebuilt a similar flow for a B2B SaaS and cut the steps from 9 to 4. I would start with a 30-minute review of your analytics. Which step do most people drop at?”

    Example session: the jobs and numbers are illustrative.

Not at your desk? Get the same jobs in Telegram

Claude finds jobs when you ask. For jobs in between, ask it to turn on Telegram alerts for your feed: new matches arrive in the bot within a minute of posting.

Ask your agent

  • Send these jobs to Telegram too: turn on alerts for my flconsole feed

Then let Claude keep that feed accurate: the feed tuner checks it on a schedule and fixes the search and AI filter, with Undo for every change.

How the feed tuner works
Step by step: set it up in the Claude app

Use case 2

Developer in Claude Code: a loop every 15 minutes, tuned by numbers

Dan builds React apps for fintech. Claude Code checks new jobs while he codes, and once a week it shows the numbers and suggests one change to the search.

Who
Developer
Goal
See fitting jobs soon after posting
Works in
Claude Code on two computers
Way
Feeds for your agent + memory
  1. Step 1: Add the skill and set up the search

    The upwork-job-hunter skill walks Claude Code through the 3-step setup and saves the search on flconsole.

    npx skills add droopboop/flconsole-upwork-mcp

    Then ask Claude Code

    • Set up my Upwork job search
  2. Step 2: Run it on a loop

    Each pass returns only jobs posted since the last check.

    Ask your agent

    • /loop 15m check new Upwork jobs in my flconsole feeds and show only the ones that fit
  3. Step 3: Once a week, tune it with numbers

    Ask your agent

    • Review my Upwork search: show the numbers for the last 7 days and suggest one change
    Your AI agent

    Review my Upwork search: show the numbers for the last 7 days and suggest one change

    flconsole · preview_filtersabout 11 jobs a day over 7 days

    Your search "react fintech" this week:

    Checked
    74 new jobs
    Fit
    9
    Top skip
    WordPress (21)
    Change
    add NOT wordpress: about 8 a day

    yes

    flconsole · update_feed"react fintech" updated

    Example session: the numbers are illustrative. Your agent keeps the tally in the feed's notes.

Same search on your work computer

Connect flconsole there too and ask the line below. Your feed, its tally and the agent's notes about you come from the server, not from a local file.

Ask your agent

  • Pick up my Upwork job search from flconsole
Step by step: run it on a loop

Use case 3

Agency and team: one search per specialization, jobs in the team chat

A small agency has three teams. The lead sets the searches up once with the agent; each team gets its own jobs in its own Telegram chat, and the lead reviews all of them in one place.

Who
Agency lead and team
Goal
The right jobs to the right people
Works in
Any agent + Telegram
Way
Feeds with Telegram alerts
  1. Step 1: Ask the agent to set up a search and a feed for each team

    Ask your agent: paste this prompt
    We are an agency with three teams: n8n automation, Shopify apps and React.
    Save a flconsole search for each, with the client filters we use everywhere: payment verified, spent $5K+, hourly from $35 or fixed from $1,000.
    Count matching jobs for each with preview_filters.
    Then create a Telegram feed for each, with the same filters and our criteria as the AI filter. Show me each feed before you create it.
  2. Step 2: Post each feed to its team chat

    In the flconsole bot, send /channel, tap Add to a group and pick the team's chat. The bot asks which feed to post there. Repeat for each team.

    Example: the counts and the job are illustrative.

    Your AI agent

    flconsole · create_feed × 33 feeds, alerts off

    flconsole · preview_filtersn8n ~6, shopify apps ~4, react ~5 a day

    Turn on Telegram alerts for "shopify apps"? New matching jobs will come to your Telegram.

    yes

    flconsole · set_feed_alertsAlerts on for "shopify apps"

    Feed posted to the team chat

    Shopify teamgroup · flconsole bot

    👋 This chat will now get new Upwork jobs for "shopify apps", right after they're posted.

    posted 1 minute ago

    ElevenLabs voice agent for a GoHighLevel agency (inbound calls + booking)

    ⭐️ 4.8 · United States

    Spent $41.6K · Hire rate 71%

    $45-70/hr · Expert

    🔔 Get jobs like this, free
  3. Step 3: Review all teams in one place

    Each job says which feed it matched, so your agent can sort them by team.

    Ask your agent

    • Check new jobs in all our flconsole feeds, group them by feed and draft an outline for the best one in each
Step by step: feeds with Telegram alerts

Use case 4

Pick a niche: compare demand over 7 days before you commit

Sara knows automation and wants to pick one tool to sell on Upwork. Her agent counts new jobs for each niche over the last 7 days and saves the winner as a feed.

Who
New to Upwork, or adding a skill
Goal
A niche with steady demand
Works in
Any agent
Way
Search only, then a feed for the agent
  1. Step 1: Ask your agent to compare the niches

    Ask your agent: paste this prompt
    I want to pick one Upwork niche in automation.
    With flconsole preview_filters, compare n8n, Zapier, Make.com and AI agent development over the last 7 days: about how many jobs a day, and the rates or budgets in the samples.
    Put it in a table and tell me where to start. Then save a search for the niche I pick.
  2. Step 2: Compare, pick, save

    Your AI agent

    Compare n8n, Zapier, Make.com and AI agent development

    flconsole · preview_filters × 47 days of jobs for each niche

    About how many new jobs a day:

    n8n
    ~15
    Zapier
    ~13
    Make.com
    ~9
    AI Agent Development
    ~30

    AI agent development has the most jobs. n8n has fewer but steady ones, and it is closest to what you do. Start there?

    n8n

    flconsole · create_feed"n8n" saved, alerts off

    Example session. The jobs-a-day numbers match flconsole data in the table below.

The real numbers behind it

New Upwork jobs tagged with each skill, 30 days to 21 October 2026.
NicheJobs a dayHourly, medianFixed, medianPayment verified
n8n~15$26/hr$10068%
Zapier~13$26/hr$10072%
Make.com~9$25/hr$12568%
AI Agent Development~30$30/hr$12578%
Source: flconsole data, 30 days to 21 October 2026. Rates and budgets are what clients offered, median. A job counts for every skill it is tagged with. More: n8n jobs on Upwork, rates by skill.
Step by step: your agent finds jobs that fit

Questions

Your AI agent gets new Upwork jobs by your search, job and client filters and keeps the ones that fit you. You can ask for jobs in a chat, check them on a loop, save searches and notes on the server, turn a search into Telegram alerts, post a feed to a team chat, or compare niches by job counts.

No. In the Claude app you add flconsole as a custom connector and paste a ready prompt into a chat. Claude asks three short questions and then shows the jobs that fit. Set it up in the Claude app

Yes. Your feeds and the agent's notes about you are kept on flconsole. Any agent you connect calls get_agent_context first and continues the same work.

Yes. Your agent creates a feed for each specialization, and in the flconsole bot you add it to your team's Telegram group with /channel. New jobs from that feed are posted there.

Your agent drafts proposals for the jobs you pick. To send them, use Upwork's MCP: it sends a proposal only after your OK. flconsole + Upwork's MCP

Yes, and it stays free: no API key, no card, no paid plan. The MCP server, feeds, Telegram alerts and the AI filter are all free.

Yes. flconsole doesn't need access to your Upwork profile: it never asks for your Upwork login, never connects to your Upwork account and never sends anything in your name.

Connect your agent and try one

Claude Code

claude mcp add --transport http flconsole https://mcp.lite.flconsole.com/mcp

Then run /mcp → Authenticate. You'll confirm in Telegram — no password, no token.

Codex

codex mcp add flconsole --url https://mcp.lite.flconsole.com/mcpcodex mcp login flconsole

A browser opens — confirm in Telegram and you're in.

Claude app

Customize → Connectors → + → Add custom connector

https://mcp.lite.flconsole.com/mcp

Works on web, desktop and mobile. You'll confirm in Telegram.

Cursor

{  "mcpServers": {    "flconsole": {      "url": "https://mcp.lite.flconsole.com/mcp"    }  }}

Add it to ~/.cursor/mcp.json, then sign in when Cursor asks.

Free · Sign in with Telegram · No API key