.claude/settings.local.json

@@ -0,0 +1,21 @@
+{
+  "permissions": {
+    "allow": [
+      "Bash(git push:*)",
+      "Read(//tmp/bluesky-collector/**)",
+      "Bash(mkdir -p \"/Users/pieter/Nextcloud-Hetzner/PXS Cloud/Projects/26004 HEIO 2/04 Applications/mastodon-collector/app/analyzer\")",
+      "Bash(docker-compose build)",
+      "Bash(docker compose build)",
+      "Bash(docker compose up -d)",
+      "Bash(docker exec mastodon-collector-collector-1 bash -c \"ANALYZER_LIMIT=100 python -m app.analyzer\")",
+      "Bash(docker compose build collector)",
+      "Bash(docker compose up -d collector)",
+      "Bash(docker compose build web)",
+      "Bash(docker compose up -d web)",
+      "Bash(curl -s -o /dev/null -w \"%{http_code}\" http://localhost:8585/analysis)",
+      "Bash(docker logs mastodon-collector-web-1 --tail 30)"
+    ],
+    "deny": [],
+    "ask": []
+  }
+}

.gitignore

@@ -48,9 +48,6 @@ venv.bak/
 .DS_Store
 .claude/
 
-# Local documentation
-LOCAL_OPERATIONS.md
-
 # Database files
 *.sqlite
 *.sqlite3

README.md

@@ -1,6 +1,6 @@
 # Mastodon Collector
 
-Collects posts, replies, and mentions from a list of Mastodon accounts and stores them in PostgreSQL. Includes automated toxicity analysis using LLM API, a web UI for account management, data browsing, and manual review of flagged content, plus JSON/CSV APIs for your analysis pipeline.
+Collects posts, replies, and mentions from a list of Mastodon accounts and stores them in PostgreSQL. Includes automated toxicity analysis using LLMs, a web UI for account management, data browsing, and manual review of flagged content, plus JSON/CSV APIs for your analysis pipeline.
 
 ## Quick Start
 
@@ -38,7 +38,7 @@ Edit `.env` to customize:
 POSTGRES_PASSWORD=collector_secret      # Change for production
 FLASK_SECRET_KEY=change-me-in-production
 POLL_INTERVAL_SECONDS=14400             # Default: 4 hours (14400s)
-LLM_API_KEY=sk-...                      # Required for toxicity analysis
+OPENAI_API_KEY=sk-...                   # Required for toxicity analysis
 ```
 
 ## Toxicity Analysis
@@ -47,7 +47,7 @@ The system includes automated toxicity detection and manual review capabilities:
 
 ### Features
 
-- **Automated Classification**: Uses LLM API to analyze posts across 12 toxicity dimensions:
+- **Automated Classification**: Uses an LLM to analyze posts across 12 toxicity dimensions:
   - General toxicity, threats, hate speech
   - Racism, antisemitism, islamophobia
   - Sexism, homophobia, ableism

accounts.txt

@@ -0,0 +1,100 @@
+# Tweede Kamer
+@HenriBontenbalCDA@mastodon.social
+@laurensdassen@mastodon.nl
+@barbarakathmann@mstdn.social
+@JesseKlaver@mstdn.social
+@JanPaternotte@mastodon.online
+@katipiri@respublicae.eu
+@Annemarijke@mastodon.social
+@LisaWesterveld@mastodon.social
+
+# Eerste Kamer
+@daanroovers@mastodon.social
+
+# Partijen landelijk
+@BIJ1@social.bij1.org
+@D66@mastodon.social
+@PartijvoordeDieren@mastodon.social
+@Piratenpartij@mastodon.social
+@voltnederland@mastodon.nl
+
+# Kabinet
+@MinisterBZK@social.overheid.nl
+@staatssecretarisbzk@social.overheid.nl
+
+# Raadsleden & wethouders
+@elisabethijmker@amsterdam.nl
+@onno@waag.social
+@erikwesselius@mastodon.social
+@paullieverse@mastodon.nl
+@ErikJonker@mastodon.social
+@tvanelferen@mastodon.social
+@joepbc@mastodon.social
+@lynchantropen@mastodon.social
+@MirjamHubert@mastodon.nl
+
+# Partijen gemeentelijk
+@PvdDAlkmaar@mastodon.social
+@ArnhemNijmegenBIJ1@social.bij1.org
+@GroenLinks026@mastodon.nl
+@pga_asten@mastodon.nl
+@groenlinks020@mastodon.social
+@pvddamsterdam@mastodon.social
+@d66debilt@mastodon.social
+@D66Bunnik@mastodon.online
+@ppdelft@mastodon.pirateparty.be
+@PvdDDelft@mastodon.social
+@PvdDDenBosch@mastodon.nl
+@DenHaagBIJ1@social.bij1.org
+@voltenschede@tukkers.online
+@PRO_Heeze_Leende@mastodon.nl
+@groenlinkshengelo@mastodon.social
+@Groenlinkspvdahouten@mastodon.social
+@volthouten@mastodon.social
+@GroenLinksPvdAKampen@mastodon.nl
+@D66Nijmegen@mastodon.nl
+@glnijmegen@mastodon.nl
+@PvdDNijmegen@mastodon.nl
+@glmeppel@mastodon.nl
+@D66Ooststellingwerf@mastodon.social
+@PvdAGLDRV@mastodon.nl
+@UtrechtBIJ1@social.bij1.org
+@d66vught@mastodon.nl
+@ProgressiefWoerden@mastodon.nl
+
+# Gemeentelijk overig
+@WheelieNick@mastodon.nl
+@hotelbreakfast@mastodon.social
+@johannesbeers@mastodon.social
+@TiciaVerveer@mastodon.social
+@RZondervan@mastodon.green
+
+# Partijen provinciaal
+@D66Brabant@mastodon.nl
+@Statenfractie_PvdD_Drenthe@mastodon.social
+@PvdDStatenfractie_Fryslan@mastodon.social
+@GroenLinksPU@mastodon.nl
+
+# Statenleden
+@SiskaPeeks@mstdn.social
+@Marjolein@mastodon.social
+
+# Europees Parlement
+@bartgroothuis@mastodon.online
+@kimvsparrentak@eupolicy.social
+
+# Waterschappen leden
+@veerleslegers@mastodon.social
+@PensioNien@todon.nl
+@fabianzoon@mastodon.nl
+@win_scheijde@mastodon.social
+@Matthijs85@mastodon.social
+
+# Waterschappen partijen
+@PvdDHHNK@mastodon.nl
+
+# Overige politici
+@alexandravanhuffelen@mastodon.social
+
+# Oud-Tweede Kamerleden
+@mariekekoekkoek@mastodon.nl

app/analyzer/analyzer.py

@@ -1,7 +1,7 @@
 """Main toxicity analysis orchestrator.
 
 Runs as a one-shot batch process: fetches unscored statuses,
-classifies them in batches with LLM API, and stores scores in PostgreSQL.
+classifies them in batches with GPT-4o-mini, and stores scores in PostgreSQL.
 
 Usage:
     python -m app.analyzer
@@ -120,7 +120,7 @@ async def run() -> None:
 
     db = AnalyzerDB(config.database_url)
     classifier = ToxicityClassifier(
-        api_key=config.llm_api_key,
+        api_key=config.openai_api_key,
         model=config.model,
     )
 

app/analyzer/classifier.py

@@ -1,6 +1,6 @@
-"""LLM-powered toxicity classifier.
+"""OpenAI-powered toxicity classifier.
 
-Sends batches of Mastodon statuses to LLM API for multi-category toxicity scoring.
+Sends batches of Mastodon statuses to GPT-4o-mini for multi-category toxicity scoring.
 Returns a list of dicts of category → score (0.0–1.0).
 """
 
@@ -181,7 +181,7 @@ def parse_batch_response(raw: str, batch_size: int) -> list[ToxicityScores]:
 
 
 class ToxicityClassifier:
-    """Async LLM-based toxicity classifier with batch support."""
+    """Async OpenAI-based toxicity classifier with batch support."""
 
     def __init__(self, api_key: str, model: str = "gpt-4o-mini"):
         self.client = AsyncOpenAI(api_key=api_key)

app/analyzer/config.py

@@ -11,7 +11,7 @@ class AnalyzerConfig:
     """Configuration for the toxicity analyzer."""
 
     database_url: str
-    llm_api_key: str
+    openai_api_key: str
     model: str = "gpt-4o-mini"
     batch_size: int = 10
     concurrency: int = 5
@@ -28,7 +28,7 @@ class AnalyzerConfig:
         """Load configuration from environment variables."""
         return cls(
             database_url=os.environ["DATABASE_URL"],
-            llm_api_key=os.environ["LLM_API_KEY"],
+            openai_api_key=os.environ["OPENAI_API_KEY"],
             model=os.getenv("ANALYZER_MODEL", "gpt-4o-mini"),
             batch_size=int(os.getenv("ANALYZER_BATCH_SIZE", "10")),
             concurrency=int(os.getenv("ANALYZER_CONCURRENCY", "5")),

docker-compose.yml

@@ -40,7 +40,7 @@ services:
     environment:
       DATABASE_URL: postgresql://collector:${POSTGRES_PASSWORD:-collector_secret}@db:5432/mastodon_collector
       POLL_INTERVAL_SECONDS: ${POLL_INTERVAL_SECONDS:-14400}
-      LLM_API_KEY: ${LLM_API_KEY}
+      OPENAI_API_KEY: ${OPENAI_API_KEY}
     volumes:
       - ./accounts.txt:/app/accounts.txt
     depends_on:

requirements.txt

@@ -5,5 +5,5 @@ sqlalchemy==2.0.36
 requests==2.32.3
 apscheduler==3.10.4
 beautifulsoup4==4.12.3
-openai==1.58.1  # OpenAI-compatible API client (supports any LLM provider)
+openai==1.58.1
 asyncpg==0.30.0