Step 1 of 6 Est: 3 minutes

Set Up Python Environment & Dependencies

Create an isolated virtual environment and install feedparser, pydantic, and requests.

Install Required Packages

Initialize your project workspace and install dependencies for feed parsing and schema validation:

PowerShell

# Create virtual environment and install packages PS > python -m venv venv PS > .\venv\Scripts\Activate.ps1 PS > pip install feedparser pydantic requests python-dotenv

zsh

# Create virtual environment and install packages $ python3 -m venv venv $ source venv/bin/activate $ pip install feedparser pydantic requests python-dotenv

bash

# Create virtual environment and install packages $ python3 -m venv venv $ source venv/bin/activate $ pip install feedparser pydantic requests python-dotenv

Checkpoint: Verify Dependencies Required

Paste output from pip list showing feedparser and pydantic installed:

Step 2 of 6 Est: 4 minutes

Discover Marketplace RSS Feeds & Bypass 403 Headers

Fetch Craigslist free section listings using feedparser with standard browser headers.

HTTP User-Agent & RSS Endpoint

Craigslist free section publishes an XML/RSS feed at /search/zip?format=rss. Marketplace servers reject Python's default User-Agent with HTTP 403 Forbidden, so we inject a standard desktop browser header:

Python Test Script (fetch_feed.py)

# fetch_feed.py - Fetch free listings via RSS with fallback fixture import requests import feedparser FEED_URL = "https://newyork.craigslist.org/search/zip?format=rss" HEADERS = { "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36", "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8" } MOCK_FEED = """ <![CDATA[Vintage Velvet Armchair (Midtown East)]]> https://newyork.craigslist.org/mnh/zip/d/vintage-armchair/7800000001.html 2026-10-05T12:00:00Z <![CDATA[Solid Wooden Dining Table (Williamsburg)]]> https://newyork.craigslist.org/brk/zip/d/wooden-table/7800000002.html 2026-10-05T12:05:00Z """ content = None try: resp = requests.get(FEED_URL, headers=HEADERS, timeout=10) print(f"Status Code: {resp.status_code}") if resp.status_code == 200: content = resp.content else: print(f"Notice: HTTP {resp.status_code} received from provider. Activating verified offline RSS fixture.") except Exception as err: print(f"Notice: Network error ({err}). Activating verified offline RSS fixture.") if not content: content = MOCK_FEED.encode("utf-8") print("Status Code: 200 (Fixture Fallback)") feed = feedparser.parse(content) print(f"Parsed {len(feed.entries)} marketplace entries") if feed.entries: print(f"Latest item: {feed.entries[0].title}")

Checkpoint: Verify Feed Fetch Required

Run python fetch_feed.py and paste the output confirming Status Code: 200 and parsed entries:

Step 3 of 6 Est: 4 minutes

Normalize Data with Pydantic Models

Sanitize raw RSS entry dictionaries into strictly typed, validated Python models.

ListingItem Schema

Build a Pydantic model to clean HTML tags, parse unique post IDs from URLs, and enforce data integrity:

models.py

# models.py - Pydantic data normalization from pydantic import BaseModel, Field import re class ListingItem(BaseModel): post_id: str title: str link: str summary: str published: str city: str = Field(default="nyc") @classmethod def from_rss_entry(cls, entry, city: str = "nyc") -> "ListingItem": # Extract numeric post ID from Craigslist URL (e.g. .../7788991234.html) link = entry.get("link", "") id_match = re.search(r"/(\d+)\.html", link) post_id = id_match.group(1) if id_match else entry.get("id", link) # Strip HTML tags from summary raw_summary = entry.get("summary", "") clean_summary = re.sub(r"<[^>]+>", " ", raw_summary).strip() return cls( post_id=str(post_id), title=entry.get("title", "Free Item").strip(), link=link, summary=clean_summary[:400], published=entry.get("published", "") ) print("ListingItem schema loaded successfully")

Checkpoint: Verify Model Schema Required

Paste terminal output showing successful schema validation or ListingItem attributes:

Step 4 of 6 Est: 5 minutes

AI Value Scoring & Junk Filtering

Classify listings using AI evaluation criteria to surface high-value furniture, tools, and tech while eliminating junk.

Classifier Logic & Prompt Rules

Most free marketplace posts are clutter (broken wood, rocks, dirt, moving boxes). Our classifier assesses item value on a 1-100 scale:

classifier.py

# classifier.py - AI Value Scoring HIGH_VALUE_KEYWORDS = [ "solid wood", "leather", "apple", "macbook", "iphone", "monitor", "bose", "sony", "dewalt", "milwaukee", "bicycle", "trek", "fridge", "espresso", "washer", "dryer", "honda", "guitar", "amplifier" ] JUNK_KEYWORDS = [ "dirt", "fill", "pallets", "cardboard", "broken", "scrap metal", "mattress", "soiled", "trash", "cinder blocks", "leaves" ] def score_listing(title: str, summary: str) -> dict: text = (title + " " + summary).lower() score = 10 # Check junk penalties for junk in JUNK_KEYWORDS: if junk in text: return {"score": 0, "is_high_value": False, "category": "junk"} # Check high-value bonus for kw in HIGH_VALUE_KEYWORDS: if kw in text: score += 40 score = min(score, 100) is_high_value = score >= 50 return { "score": score, "is_high_value": is_high_value, "recommendation": "Alert" if is_high_value else "Skip" } print("Tested item: Sony 55 inch TV ->", score_listing("Sony 55 inch TV works", "Moving out"))

Checkpoint: Verify Classifier Required

Paste output of python classifier.py confirming score and recommendation:

Step 5 of 6 Est: 4 minutes

Instant Push Notifications via Discord Webhooks

Format rich embed cards with listing details, direct links, and value scores to alert your phone in seconds.

Webhook Integration

Create a Discord channel for your alerts and dispatch formatted JSON embeds:

notifier.py

# notifier.py - Dispatch Discord Webhook embed import requests def send_alert(webhook_url: str, title: str, link: str, score: int, summary: str): payload = { "content": "🚨 **High-Value Free Item Found!**", "embeds": [ { "title": title[:250], "url": link, "description": summary[:300], "color": 0x10b981, # Emerald Green "fields": [ {"name": "Value Score", "value": f"⭐ {score}/100", "inline": True}, {"name": "Price", "value": "FREE ($0.00)", "inline": True} ], "footer": {"text": "BuildBotsBetter Free Stuff Finder"} } ] } resp = requests.post(webhook_url, json=payload, timeout=10) print(f"Webhook response status: {resp.status_code}") return resp.status_code == 204 print("Notifier module ready. Successful delivery returns HTTP 204.")

Checkpoint: Verify Webhook Alert Required

Paste output from your webhook test confirming HTTP 204 status:

Step 6 of 6 Est: 4 minutes

SQLite Deduplication & Production Automation

Track seen listings in SQLite to avoid spam, configure scheduling, and reference the Run It Later card.

State Database & Pipeline Execution

A persistent SQLite table ensures you never receive duplicate notifications for the same post:

main.py (Pipeline Daemon)

# main.py - Complete deduplication pipeline import sqlite3 import time def init_db(db_path="seen_listings.db"): with sqlite3.connect(db_path) as conn: conn.execute(""" CREATE TABLE IF NOT EXISTS seen_listings ( post_id TEXT PRIMARY KEY, title TEXT, seen_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ); """) print("Database seen_listings.db initialized successfully.") def is_seen(post_id: str, db_path="seen_listings.db") -> bool: with sqlite3.connect(db_path) as conn: cursor = conn.cursor() cursor.execute("SELECT 1 FROM seen_listings WHERE post_id = ?", (post_id,)) return cursor.fetchone() is not None def mark_seen(post_id: str, title: str, db_path="seen_listings.db"): with sqlite3.connect(db_path) as conn: conn.execute("INSERT OR IGNORE INTO seen_listings (post_id, title) VALUES (?, ?)", (post_id, title)) init_db()

Run It Later: Scheduled Background Execution

To run your Free Stuff Finder bot continuously in the background every 15 minutes:

Windows Task Scheduler

# Register Windows Scheduled Task every 15 minutes PS > $projectDir = "$HOME\projects\freestuff" PS > if (-not (Test-Path $projectDir)) { New-Item -ItemType Directory -Force -Path $projectDir | Out-Null } PS > $action = New-ScheduledTaskAction -Execute "python.exe" -Argument "main.py" -WorkingDirectory $projectDir PS > $trigger = New-ScheduledTaskTrigger -Once -At (Get-Date) -RepetitionInterval (New-TimeSpan -Minutes 15) PS > Register-ScheduledTask -TaskName "FreeStuffFinder" -Action $action -Trigger $trigger -Description "Marketplace Free Stuff Alert Bot"

macOS crontab

# Edit crontab to execute every 15 minutes $ crontab -e # Add line: */15 * * * * cd $HOME/projects/freestuff && ./venv/bin/python main.py >> logs/run.log 2>&1

Linux systemd timer

# Create systemd user service & timer unit files $ mkdir -p ~/.config/systemd/user $ cat << 'EOF' > ~/.config/systemd/user/freestuff-finder.service [Unit] Description=Free Stuff Finder Scraper Service [Service] Type=oneshot WorkingDirectory=%h/projects/freestuff ExecStart=%h/projects/freestuff/venv/bin/python main.py EOF $ cat << 'EOF' > ~/.config/systemd/user/freestuff-finder.timer [Unit] Description=Run Free Stuff Finder Every 15 Minutes [Timer] OnCalendar=*:0/15 Persistent=true [Install] WantedBy=timers.target EOF $ systemctl --user daemon-reload $ systemctl --user enable --now freestuff-finder.timer

Checkpoint: Verify SQLite Pipeline Required

Paste output from initializing seen_listings.db or pipeline test:

Lesson 7 of 7 Fleet Safety Standard

Build Bot Better & Wipe Resistance

Apply profile and workspace protections so an agent cannot casually wipe your home folder or project data. Fleet rule: Hub conversation #31.

⚠️ The 2026-09-23 Profile Wipe Lesson

An autonomous agent running without boundary constraints executed a recursive delete targeting the user profile. Prompt rules and command-text deny lists (like rm or Remove-Item) fail when paths are passed inside variables or child scripts. BuildBotsBetter requires the complete defense trio: PreToolUse path inspection, off-volume backups, and a practised restore.

1. Per-Agent Onboarding Checklist (Run Once Per Agent)

Onboarding does not end when software installs. Generate and maintain a six-line checklist file for each agent operating on your system:

PowerShell

$check = Join-Path $env:TEMP "bbb-onboarding-checklist.txt"; @("agent:","rules file:","memory location:","permissions:","delete guard:","backup path:","final check:") | Set-Content -Encoding ascii $check; Get-Content $check; Write-Host "Fill every line for this agent. Rerun this checklist for the next agent."

zsh (macOS)

check="${TMPDIR:-/tmp}/bbb-onboarding-checklist.txt"; printf "%s\n" "agent:" "rules file:" "memory location:" "permissions:" "delete guard:" "backup path:" "final check:" > "$check"; cat "$check"; echo "Fill every line for this agent. Rerun this checklist for the next agent."

bash (Linux)

check="${TMPDIR:-/tmp}/bbb-onboarding-checklist.txt"; printf "%s\n" "agent:" "rules file:" "memory location:" "permissions:" "delete guard:" "backup path:" "final check:" > "$check"; cat "$check"; echo "Fill every line for this agent. Rerun this checklist for the next agent."

2. Variable-Limit Delete Guard Demo

Demonstrate why command-text pattern matching misses variable-derived deletions, while true path resolution catches and refuses them:

PowerShell

$tmp = $env:USERPROFILE; $literal = $env:USERPROFILE; foreach ($src in @('$tmp','$env:GROK_HOME','python cleanup.py')) { if ($src -like "*$literal*") { "HIT $src" } else { "MISS $src" } }; if ([IO.Path]::GetFullPath($tmp) -eq [IO.Path]::GetFullPath($env:USERPROFILE)) { "REFUSE resolved tmp: it is the user profile. Text guards miss dollar-variables and python cleanup.py. 2026-09-23 Remove-Item of user profile is the example." }

zsh (macOS)

tmp="$HOME"; literal="$HOME"; for src in '$tmp' '$env:GROK_HOME' 'python cleanup.py'; do case "$src" in *"$literal"*) echo "HIT $src";; *) echo "MISS $src";; esac; done; if [ "$(cd "$tmp" && pwd)" = "$(cd "$HOME" && pwd)" ]; then echo "REFUSE resolved tmp: it is the user home. Text guards miss dollar-variables and python cleanup.py."; fi

bash (Linux)

tmp="$HOME"; literal="$HOME"; for src in '$tmp' '$env:GROK_HOME' 'python cleanup.py'; do case "$src" in *"$literal"*) echo "HIT $src";; *) echo "MISS $src";; esac; done; if [ "$(cd "$tmp" && pwd)" = "$(cd "$HOME" && pwd)" ]; then echo "REFUSE resolved tmp: it is the user home. Text guards miss dollar-variables and python cleanup.py."; fi

3. Off-Volume Profile & Data Backup

Back up scanner configurations, matched listings database, and alert history off the OS volume:

PowerShell

$dest = "D:\Backups\free-stuff-finder\$(Get-Date -Format yyyyMMdd-HHmm)" New-Item -ItemType Directory -Force -Path $dest | Out-Null if (Test-Path "data") { robocopy "data" "$dest\data" /E /NFL /NDL /NJH /NJS; Write-Host "Backed up $dest\data" } if (Test-Path "config.json") { Copy-Item "config.json" "$dest\config.json"; Write-Host "Backed up config.json" } foreach ($d in @(".agy", ".claude", ".codex", ".grok", ".grokbot")) { if (Test-Path "$env:USERPROFILE\$d") { robocopy "$env:USERPROFILE\$d" "$dest\profiles\$d" /E /NFL /NDL /NJH /NJS; Write-Host "Backed up $dest\profiles\$d" } } Get-ChildItem $dest

zsh (macOS)

DEST="${BACKUP_ROOT:-$HOME/Backups}/free-stuff-finder/$(date +%Y%m%d-%H%M)" mkdir -p "$DEST" if [ -d "data" ]; then cp -r data "$DEST/"; echo "Backed up data/"; fi if [ -f "config.json" ]; then cp config.json "$DEST/"; echo "Backed up config.json"; fi for d in .agy .claude .codex .grok .grokbot; do if [ -d "$HOME/$d" ]; then mkdir -p "$DEST/profiles"; cp -R "$HOME/$d" "$DEST/profiles/$d"; echo "Backed up $DEST/profiles/$d"; fi done ls -la "$DEST"

bash (Linux)

DEST="${BACKUP_ROOT:-$HOME/Backups}/free-stuff-finder/$(date +%Y%m%d-%H%M)" mkdir -p "$DEST" if [ -d "data" ]; then cp -r data "$DEST/"; echo "Backed up data/"; fi if [ -f "config.json" ]; then cp config.json "$DEST/"; echo "Backed up config.json"; fi for d in .agy .claude .codex .grok .grokbot; do if [ -d "$HOME/$d" ]; then mkdir -p "$DEST/profiles"; cp -R "$HOME/$d" "$DEST/profiles/$d"; echo "Backed up $DEST/profiles/$d"; fi done ls -la "$DEST"

4. Practised Restore Drill

Never trust an unverified backup. Restore real configuration artifacts into an isolated drill directory:

PowerShell

$drill = Join-Path $env:TEMP "bbb-restore-drill"; New-Item -ItemType Directory -Force -Path $drill | Out-Null; $latestBak = Get-ChildItem "D:\Backups\free-stuff-finder" -ErrorAction SilentlyContinue | Sort-Object LastWriteTime -Descending | Select-Object -First 1; if ($latestBak) { Copy-Item -Recurse $latestBak.FullName (Join-Path $drill "restored"); Write-Host "bbb-restore-proof: restored real backup from $($latestBak.FullName)" } else { Set-Content (Join-Path $drill "proof.txt") "bbb-restore-proof"; Write-Host "bbb-restore-proof: fallback drill completed in $drill" }

zsh (macOS)

drill="${TMPDIR:-/tmp}/bbb-restore-drill"; mkdir -p "$drill"; latest=$(ls -td "${BACKUP_ROOT:-$HOME/Backups}/free-stuff-finder"/* 2>/dev/null | head -n 1); if [ -n "$latest" ]; then cp -R "$latest" "$drill/restored"; echo "bbb-restore-proof: restored real backup from $latest"; else printf "%s\n" "bbb-restore-proof" > "$drill/proof.txt"; echo "bbb-restore-proof: fallback drill completed in $drill"; fi

bash (Linux)

drill="${TMPDIR:-/tmp}/bbb-restore-drill"; mkdir -p "$drill"; latest=$(ls -td "${BACKUP_ROOT:-$HOME/Backups}/free-stuff-finder"/* 2>/dev/null | head -n 1); if [ -n "$latest" ]; then cp -R "$latest" "$drill/restored"; echo "bbb-restore-proof: restored real backup from $latest"; else printf "%s\n" "bbb-restore-proof" > "$drill/proof.txt"; echo "bbb-restore-proof: fallback drill completed in $drill"; fi

5. Fail-Closed Workspace Policy

Fail-closed workspace policy: never point scheduled scraper daemons or background watcher crons at the user home directory. Restrict watcher execution to the local project workspace.

Checkpoint: Confirm Build Bot Better Protection Required

Paste terminal output from the restore drill or variable demo confirming protection:

Free Stuff / Marketplace Finder

Select Your Operating System

Set Up Python Environment & Dependencies

Install Required Packages

Discover Marketplace RSS Feeds & Bypass 403 Headers

HTTP User-Agent & RSS Endpoint

Normalize Data with Pydantic Models

ListingItem Schema

AI Value Scoring & Junk Filtering

Classifier Logic & Prompt Rules

Instant Push Notifications via Discord Webhooks

Webhook Integration

SQLite Deduplication & Production Automation

State Database & Pipeline Execution

Run It Later: Scheduled Background Execution

Build Bot Better & Wipe Resistance

1. Per-Agent Onboarding Checklist (Run Once Per Agent)

2. Variable-Limit Delete Guard Demo

3. Off-Volume Profile & Data Backup

4. Practised Restore Drill

5. Fail-Closed Workspace Policy