A tool for tailing a labelers' firehose, rehydrating, and storing records for future analysis of moderation decisions.

Update README.md

+6 -4
+3 -1
.claude/settings.local.json
··· 2 2 "permissions": { 3 3 "allow": [ 4 4 "Bash(git add:*)", 5 - "Bash(git commit:*)" 5 + "Bash(git commit:*)", 6 + "mcp__git-mcp-server__git_push", 7 + "mcp__git-mcp-server__git_log" 6 8 ], 7 9 "deny": [], 8 10 "ask": []
+3 -3
README.md
··· 68 68 WSS_URL=wss://your-labeler.com/xrpc/com.atproto.label.subscribeLabels 69 69 70 70 # Optional: Filter specific labels 71 - CAPTURE_LABELS=spam,hate-speech,csam 71 + CAPTURE_LABELS=spam,hate-speech 72 72 73 73 # Logging 74 74 LOG_LEVEL=info ··· 173 173 Filter labels by providing a comma-separated list in `CAPTURE_LABELS`: 174 174 175 175 ```env 176 - CAPTURE_LABELS=spam,hate-speech,csam,scam 176 + CAPTURE_LABELS=spam,hate-speech,scam 177 177 ``` 178 178 179 179 If not set, all labels are captured. ··· 271 271 ## Safety Features 272 272 273 273 ### Blob Hydration 274 - By default, `HYDRATE_BLOBS` is `false`. This prevents accidental download of potentially harmful content (CSAM, graphic violence, etc.) while still capturing cryptographic and perceptual hashes for ML training. 274 + By default, `HYDRATE_BLOBS` is `false`. This prevents accidental download of potentially harmful / and or unlawful content (CSAM, graphic violence, etc.) while still capturing cryptographic and perceptual hashes. 275 275 276 276 Only enable blob download if: 277 277 1. You understand the legal and safety implications