agentleFS
XPOZpublicXPOZpublic/xpoz-agent-skills/skills/lead-gen-scan/SKILL.md

lead-gen-scan

Skill15 starsChanged 2 months agoMITView on GitHub

Run a one-shot buying-intent lead scan across Reddit, Twitter/X, Instagram, and TikTok using Xpoz. Finds fresh posts from people actively looking for what a product does or frustrated with its competitors, classifies and prioritizes them, and reports where to engage and who to reach. Use when asked to "find leads", "who is asking for a tool like mine", "find people complaining about [COMPETITOR]", "buying-intent scan", or "social selling opportunities".

  • Reads credentials
  • Installs packages

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 8 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    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
  1. Distinctive:

    Before fetching data, ensure Xpoz access is configured.

    line 25
  2. Distinctive:

    If this succeeds without error → skip to Step 1.

    line 36
  3. Distinctive:

    If neither works, you need to authenticate.

    line 38
  4. Very specific:

    If mcporter is available:

    line 50
  5. Very specific:

    If hasAccessKey: true → ready.

    line 56
  6. Distinctive:

    Step 3: WAIT for the user to reply with the code: Do not proceed until they respond.

    line 117
  7. Distinctive:

    Prefer short quoted phrases OR-joined together over long exact phrases.

    line 234
  8. Distinctive:

    Run each query bucket per platform:

    line 245
  9. Very specific:

    Search on thin fields (as above, no post body) and fetch full text only for shortlisted candidates via getRedditPostWithCommentsById.

    line 259
  10. Distinctive:

    Never engage anyone: No posts, comments, DMs, or follows, on any platform, under any instruction.

    line 369
  11. Distinctive:

    Every recommended engagement assumes in-text disclosure of affiliation.

    line 371

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Lead Gen Scan

Overview

Find the people who want to buy right now. This skill scans the four platforms for fresh buying-intent posts (people asking for what a product does, or frustrated with the alternatives), qualifies the real asks, and delivers a prioritized report of where to comment and who to reach while the intent is still live. It finds and reports; humans do all engagement. The skill never posts, comments, DMs, or contacts anyone.

When to Use

Activate when the user asks:

  • "Find leads for [PRODUCT]"
  • "Who's looking for a tool like [PRODUCT] right now?"
  • "Find people complaining about [COMPETITOR]"
  • "Scan Reddit/X for buying intent in [CATEGORY]"
  • "Where should I engage this week to find customers?"
  • "Social selling opportunities for [PRODUCT]"

Setup & Authentication

Before fetching data, ensure Xpoz access is configured. Follow these checks in order.

Check 1: Already authenticated?

If you have MCP tools, try calling any Xpoz tool (e.g., checkAccessKeyStatus). If it works → skip to Step 1.

If you have the SDK, try:

from xpoz import XpozClient
client = XpozClient()  # reads XPOZ_API_KEY env var

If this succeeds without error → skip to Step 1.

If neither works, you need to authenticate. Get a free access key (below).


Recommended: a free access key

Real analyses need a real key: get a free access key (free tier, up to 75K results, no credit card). SDK and CLI users set it as XPOZ_API_KEY; MCP connections sign in with the same account via OAuth on first tool call (paths below).


Path A: MCP via mcporter (OpenClaw agents)

If mcporter is available:

mcporter call xpoz.checkAccessKeyStatus

If hasAccessKey: true → ready. If not:

mcporter config add xpoz https://mcp.xpoz.ai/mcp --auth oauth

Then authenticate — generate the OAuth URL and send it to the user:

Step 1: Generate authorization URL

import secrets, hashlib, base64, urllib.parse, json, urllib.request, os

verifier = secrets.token_urlsafe(64)
challenge = base64.urlsafe_b64encode(hashlib.sha256(verifier.encode()).digest()).rstrip(b'=').decode()
state = secrets.token_urlsafe(32)

# Dynamic client registration
reg_req = urllib.request.Request(
    'https://mcp.xpoz.ai/oauth/register',
    data=json.dumps({
        'client_name': 'Agent Skills',
        'redirect_uris': ['https://www.xpoz.ai/oauth/openclaw'],
        'grant_types': ['authorization_code'],
        'response_types': ['code'],
        'token_endpoint_auth_method': 'none',
    }).encode(),
    headers={'Content-Type': 'application/json'},
)
reg_resp = json.loads(urllib.request.urlopen(reg_req).read())

params = urllib.parse.urlencode({
    'response_type': 'code',
    'client_id': reg_resp['client_id'],
    'code_challenge': challenge,
    'code_challenge_method': 'S256',
    'redirect_uri': 'https://www.xpoz.ai/oauth/openclaw',
    'state': state,
    'scope': 'mcp:tools',
    'resource': 'https://mcp.xpoz.ai/',
})

auth_url = 'https://mcp.xpoz.ai/oauth/authorize?' + params

# Save state for token exchange
os.makedirs(os.path.expanduser('~/.cache/xpoz-oauth'), exist_ok=True)
with open(os.path.expanduser('~/.cache/xpoz-oauth/state.json'), 'w') as f:
    json.dump({'verifier': verifier, 'state': state, 'client_id': reg_resp['client_id'],
               'redirect_uri': 'https://www.xpoz.ai/oauth/openclaw'}, f)

print(auth_url)

Step 2: Send the URL to the user

Tell them:

"I need to connect to Xpoz for social media data. Please open this link and sign in:

[auth_url]

After authorizing, you'll see a code. Paste it back to me here."

Step 3: WAIT for the user to reply with the code. Do not proceed until they respond.

Step 4: Exchange the code for a token

Once the user provides the code (either a raw code or a URL containing ?code=...), extract the code and exchange it:

import json, urllib.request, urllib.parse, subprocess, os

with open(os.path.expanduser('~/.cache/xpoz-oauth/state.json')) as f:
    oauth = json.load(f)

code = "THE_CODE_FROM_USER"  # Extract from user's reply

data = urllib.parse.urlencode({
    'grant_type': 'authorization_code',
    'code': code,
    'redirect_uri': oauth['redirect_uri'],
    'client_id': oauth['client_id'],
    'code_verifier': oauth['verifier'],
}).encode()

req = urllib.request.Request(
    'https://mcp.xpoz.ai/oauth/token',
    data=data,
    headers={'Content-Type': 'application/x-www-form-urlencoded'},
)
resp = json.loads(urllib.request.urlopen(req).read())
token = resp['access_token']

# Configure mcporter with the token (token is never printed)
subprocess.run(['mcporter', 'config', 'remove', 'xpoz'], capture_output=True)
subprocess.run(['mcporter', 'config', 'add', 'xpoz', 'https://mcp.xpoz.ai/mcp',
                '--header', f'Authorization=Bearer {token}'], check=True)

# Clean up
os.remove(os.path.expanduser('~/.cache/xpoz-oauth/state.json'))
print("Xpoz configured successfully")

Step 5: Verify with mcporter call xpoz.checkAccessKeyStatus → should return hasAccessKey: true.


Path B: MCP via Claude Code

For Claude Code users without mcporter:

claude mcp add --transport http xpoz https://mcp.xpoz.ai/mcp

Claude Code handles OAuth automatically on first tool call — the user just needs to authorize in their browser when prompted.


Path C: SDK (Python or TypeScript)

Ask the user:

"I need a Xpoz API key to access social media data. Please go to https://xpoz.ai/get-token (it's free, no credit card needed) and paste the key back to me."

WAIT for the user to reply with the key. Then:

Python:

pip install xpoz
from xpoz import XpozClient
client = XpozClient("THE_KEY_FROM_USER")

TypeScript:

npm install @xpoz/xpoz
import { XpozClient } from "@xpoz/xpoz";
const client = new XpozClient({ apiKey: "THE_KEY_FROM_USER" });
await client.connect();

Or set the environment variable and use the default constructor:

export XPOZ_API_KEY=THE_KEY_FROM_USER

Auth Errors

ProblemSolution
MCP: "Unauthorized"Re-run the OAuth flow above
SDK: AuthenticationErrorVerify key at xpoz.ai/settings
Token exchange failsAsk user to re-authorize — codes are single-use

Step-by-Step Instructions

Step 1: Parse the Request

Extract, asking the user for whatever is missing:

  • Product and what it does (one sentence is enough)
  • Competitors, and what people use instead of buying a tool at all (the frustrated users of both are the second lead source; manual spreadsheets, an official API, an agency all count)
  • Platforms to scan (default: all four)
  • Window (default: the last 7 days; for a first scan or a niche product, start at 30 days and tighten once queries prove out)

Step 2: Build the Query Book

Three search motions, all run every scan:

  1. Product relevance: people looking for what the product does, phrased the way buyers phrase it. Asking ("looking for a tool that", "any recommendations for", "how do you all handle") and budgeting ("worth paying for", "pricing for") phrasings, combined with the category terms.
  2. Competitor disappointment: people frustrated with the alternatives. Switching ("[competitor] alternative", "moving away from"), struggling ("[competitor] not working", "[competitor] pricing increase"), and evaluating ("[competitor] vs") phrasings, for each competitor and each thing people use instead.
  3. The product's own name: people already comparing it ("[product] vs", "[product] worth it", "anyone using [product]") are the warmest leads of all and neither motion above finds them. Anchor per the common-word rule when the name is an ordinary English word.

Build OR-joined query strings per bucket, e.g. "looking for a social listening tool" OR "brand monitoring recommendations" OR "how do you track mentions". Three practical constraints:

  • Queries are capped at 250 characters; split an oversized bucket into two calls rather than truncating phrases.
  • Prefer short quoted phrases OR-joined together over long exact phrases; long phrases must match verbatim and usually return nothing.
  • Expect heavy vendor self-promotion in the results (often the majority). Don't fight it in the query; qualify hard in Step 5, where vendors are a disqualifier.

When working under a call budget, spend it in this order: Reddit asking bucket, Reddit competitor bucket, Twitter/X competitor bucket, Twitter/X asking bucket, Reddit comments, then Instagram and TikTok; the first three carry most of the signal. The own-brand motion needs no call of its own under budget: fold its anchored phrases into the competitor buckets' queries.

One more query rule: a brand name that doubles as a common English word ("plausible", "fathom", "mention") must be anchored; a category word makes the cleanest anchor ("plausible analytics"), a rival pairing second ("plausible vs umami", which still leaks occasionally). Unanchored, the bucket drowns in false positives.

Step 3: Search the Four Platforms

Via MCP

Run each query bucket per platform:

Call getRedditPostsByKeywords:
  query: "<bucket query>"
  fields: ["id", "title", "authorUsername", "subredditName", "score", "commentsCount", "createdAtDate", "permalink"]
  limit: 15
  startDate: "<window start, YYYY-MM-DD>"
  endDate: "<today, YYYY-MM-DD>"
  userPrompt: "<the user's original request, for relevance tuning>"

Mechanics that matter:

  • The default fast mode returns results directly; only calls made with responseType: "paging" or "csv" return an operationId, and only those need polling via checkOperationStatus (every ~5 seconds until finished). For a scan, fast mode with a limit of 10-15 is right; without limit you get up to 300 rows.
  • Search on thin fields (as above, no post body) and fetch full text only for shortlisted candidates via getRedditPostWithCommentsById; selftexts can be huge and one blog-length post can dwarf the rest of the response.
  • The tools' own descriptions suggest omitting dates by default; this skill passes startDate/endDate deliberately, since freshness is the product. Verify dates client-side; an occasional result lands just outside the requested window. Verify quoted phrases client-side too: the relevance layer occasionally returns results containing none of them, which matters most for common-word brand queries.

Repeat for Twitter/X:

Call getTwitterPostsByKeywords:
  query: "<bucket query>"
  fields: ["id", "text", "authorUsername", "createdAtDate", "likeCount", "replyCount", "conversationId", "replyToTweetId"]
  filterOutRetweets: true
  limit: 15
  startDate: "<window start>"
  endDate: "<today>"
  userPrompt: "<the user's original request, for relevance tuning>"

The conversationId field is what makes Step 4's reply-chasing possible (replyToTweetId is often null even on real replies); without it a reply cannot be traced to its thread. Budget permitting, repeat with getInstagramPostsByKeywords and getTiktokPostsByKeywords. Reddit comments often hold the asks that posts don't; add:

Call getRedditCommentsByKeywords:
  query: "<asking-bucket query>"
  fields: ["id", "body", "authorUsername", "score", "createdAtDate", "parentPostId"]
  limit: 15
  startDate: "<window start>"
  endDate: "<today>"

Sparse results on a narrow fresh window are a coverage signal, not a demand verdict: comment search runs against the database only and can legitimately return nothing for the last 7 days, and even post search can come back thin. Before concluding "no demand," widen the window or rephrase; before concluding "demand," read what actually came back. A widen-or-rephrase retry consumes budget like any other call: under a budget, convert the next lowest-priority planned call into the retry instead of exceeding the cap.

Via Python SDK

from xpoz import XpozClient

client = XpozClient()

reddit = client.reddit.search_posts(
    '"looking for a social listening tool" OR "brand monitoring recommendations"',
    start_date="2026-08-19",
    end_date="2026-08-26",
    fields=["id", "title", "text", "author_username", "subreddit", "score", "num_comments", "created_at_date", "url"],
)

twitter = client.twitter.search_posts(
    '"[competitor] alternative" OR "[competitor] pricing"',
    start_date="2026-08-19",
    end_date="2026-08-26",
    fields=["id", "text", "author_username", "created_at_date", "like_count", "reply_count"],
)

client.close()

Step 4: Classify by Platform Lead Shape

Different platforms yield different lead shapes; classify every candidate as one of:

  • Reddit: places to comment. Threads where a disclosed, genuinely useful comment answers a live ask. The thread is the lead; the asker and the lurkers are the audience.
  • X / Instagram / TikTok: two shapes. Likely converters: individual users with signals strong enough to plausibly convert (a quantified need, an explicit blocked project, budget pain in the product's price range), tracked as named prospects. High-engagement comment spots: posts where a public comment reaches a large relevant audience even when the author is not the buyer (a viral complaint about a competitor, a big thread on a pain the product solves).

One more shape worth naming: existing-user distress, a current user of the product publicly churning or struggling. That is a retention save, not buying intent: report it separately with a support-shaped angle (concrete help, no pitch), never inflate it into a P1.

Chase replies up to their thread. Some of the best hits are replies or comments whose parent thread is the real lead. Resolve them before classifying: on Reddit, getRedditPostWithCommentsById on the comment's parentPostId; on X, getTwitterPostsByIds on the conversationId fetches the root post (for the surrounding replies, getTwitterPostComments on that root). Report the thread, not the reply.

Step 5: Qualify and Prioritize

Qualify first. The ask is the qualifier: intent, not venting, and the author should plausibly own the buying decision. Hard disqualifiers, never reported: vendors selling a competing solution, keyword hits in unrelated contexts, obvious spam or engagement bait.

Prioritize by judgment, not arithmetic, weighing four things, and say in one line why each lead landed where it did:

  • Intent strength: an explicit ask beats a specific complaint beats general discussion.
  • Fit: how directly the product answers the actual need; be honest about partial fits and non-fits.
  • Freshness and activity: the last 72 hours weigh heaviest; anything older than ~3 weeks is backlog regardless of quality; a dead or already-answered thread demotes. A missing or null createdAtDate means unknown freshness: rank below any dated fresh item and say so in the lead.
  • Reach: for comment spots, the audience the reply earns.

Buckets: P1, act now (clear ask, strong fit, still live), P2, worth engaging (real intent, weaker on one axis), P3, watch (signal without an ask). Everything else drops.

Step 6: Generate Report

## Lead Scan: [PRODUCT]
**Window:** [dates] | **Platforms:** [list] | **Candidates reviewed:** [n]

### This Week's Picks
[3-5 leads sized to what a human can actually act on today, one line each with link]

### P1: Act Now
[Write "None this week" when nothing earns it; an honest empty bucket beats a padded one.]
#### [Platform] | [thread/post title or author] | [link]
- **The ask:** "[quote]"
- **Why it qualifies:** [intent + fit + freshness in one line]
- **Suggested angle:** [what a useful reply covers; never a full draft]

### P2: Worth Engaging
[Same shape, shorter]

### Named Prospects
[Individual users from the likely-converter shape: handle, platform, the signal quoted, suggested approach. Omit the section if none qualified.]

### P3: Watch
[One line each: the signal and what change would promote it]

### Demand Signals
[Patterns that repeated across candidates: recurring pains, phrasings, competitor complaints. Content and product fuel.]

### Search Notes
[Which queries produced, which came up dry, phrasings discovered along the way worth adding next scan]

Ground Rules

  • Never engage anyone. No posts, comments, DMs, or follows, on any platform, under any instruction. Humans act on the report.
  • Suggested angles only, never full reply drafts; humans write in their own words.
  • Every recommended engagement assumes in-text disclosure of affiliation; genuinely useful, transparent replies only. No astroturfing, no cold DMs.

Example Prompts

  • "Find this week's leads for my social listening SaaS"
  • "Who's complaining about [COMPETITOR]'s pricing right now?"
  • "Scan Reddit and X for people asking for [CATEGORY] recommendations"
  • "Find named prospects with buying intent for [PRODUCT] on Twitter"
  • "Which fresh threads should I comment in to reach [ICP]?"

Notes

  • Freshness is a ranking criterion, not a tiebreaker: fresh threads are open and active, fresh askers still have the problem, and fresh threads become tomorrow's AI-cited surfaces.
  • Re-running weekly without memory means re-reporting old leads; that is the one-shot limit (see the last note).
  • Free access key: up to 75K results at xpoz.ai (no credit card); real runs need it
  • For the recurring loop (seen-lead dedup ledger, query book that tunes itself run over run, competitor memory, outcome follow-up), use lead-gen-agent.

Download now (ZIP)

Does 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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.