Configuration Reference¶

The Data Packet is configured entirely through environment variables. No config file is required.

Resolution order¶

Config values resolve from highest to lowest priority:

1. get_config(keyword=value)    ← Python API direct override (highest)
2. CLI flags                    ← --show-name, --male-voice, etc.
3. Environment variables        ← SHOW_NAME, MALE_VOICE, etc.
4. Built-in defaults            ← hardcoded in Config dataclass (lowest)

Required¶

Required variables

The pipeline will not start without these two.

Variable	Description
`ANTHROPIC_API_KEY`	Anthropic API key for Claude script generation. Get one at console.anthropic.com.
`GCS_BUCKET_NAME`	Google Cloud Storage bucket used as intermediate storage during Long Audio Synthesis.

Google Cloud credentials¶

Service account JSON (recommended)Application Default Credentials

GOOGLE_APPLICATION_CREDENTIALS=/path/to/service-account-key.json

Mount into Docker:

-v "$(pwd)/service-account-key.json:/credentials.json:ro" \
-e GOOGLE_APPLICATION_CREDENTIALS=/credentials.json

If running on GCP (Compute Engine, Cloud Run, GKE), credentials are provided automatically by the environment — no variable needed.

AWS S3 (optional)¶

Info

If S3_BUCKET_NAME is not set, S3 upload is silently skipped and files are saved locally only.

Variable	Default	Description
`S3_BUCKET_NAME`	—	S3 bucket name
`AWS_ACCESS_KEY_ID`	—	AWS access key
`AWS_SECRET_ACCESS_KEY`	—	AWS secret key
`AWS_REGION`	`us-east-1`	AWS region

MongoDB (optional)¶

Info

If credentials are absent, deduplication is skipped. All articles are always eligible.

Variable	Default	Description
`MONGODB_USERNAME`	—	MongoDB username
`MONGODB_PASSWORD`	—	MongoDB password
`MONGODB_HOST`	`localhost`	MongoDB host
`MONGODB_PORT`	`27017`	MongoDB port
`MONGODB_DATABASE`	`the_data_packet`	Database name

Podcast content¶

Variable	Default	Description
`SHOW_NAME`	`The Data Packet`	Podcast show name, used in script and RSS feed
`MAX_ARTICLES`	`1`	Maximum articles fetched per source per run

Audio voices¶

Studio voices only

Only en-US-Studio-* voices are compatible with Long Audio Synthesis. Standard and WaveNet voices are not supported.

Variable	Default	Description
`MALE_VOICE`	`en-US-Studio-Q`	Voice for first speaker (Alex)
`FEMALE_VOICE`	`en-US-Studio-O`	Voice for second speaker (Sam)

Available Studio voices:

Voice ID	Character
`en-US-Studio-Q`	Male — warm, professional
`en-US-Studio-O`	Female — clear, engaging
`en-US-Studio-M`	Male — authoritative

Logging¶

Variable	Default	Description
`LOG_LEVEL`	`INFO`	`DEBUG` / `INFO` / `WARNING` / `ERROR`
`LOG_DIRECTORY`	`output/logs`	Directory for JSONL log files
`ENABLE_JSONL_LOGGING`	`true`	Write structured JSONL log files
`ENABLE_S3_LOG_UPLOAD`	`true`	Upload completed daily log files to S3
`LOG_UPLOAD_INTERVAL`	`3600`	Seconds between S3 log upload checks
`REMOVE_LOGS_AFTER_UPLOAD`	`false`	Delete local files after S3 upload

Grafana Loki (optional)¶

Variable	Description
`GRAFANA_LOKI_URL`	Loki push API URL (e.g. `https://loki.example.com`)
`GRAFANA_LOKI_USERNAME`	Basic auth username
`GRAFANA_LOKI_PASSWORD`	Basic auth password

Complete `.env` example¶

.env

# ── Required ─────────────────────────────────────────────────────────────────
ANTHROPIC_API_KEY=sk-ant-api03-your-key-here
GCS_BUCKET_NAME=my-tts-audio-bucket
GOOGLE_APPLICATION_CREDENTIALS=/app/service-account-key.json

# ── AWS S3 (optional) ────────────────────────────────────────────────────────
S3_BUCKET_NAME=my-podcast-hosting-bucket
AWS_ACCESS_KEY_ID=AKIA...
AWS_SECRET_ACCESS_KEY=your-aws-secret
AWS_REGION=us-east-1

# ── MongoDB (optional) ───────────────────────────────────────────────────────
MONGODB_USERNAME=admin
MONGODB_PASSWORD=your-password

# ── Podcast settings ─────────────────────────────────────────────────────────
SHOW_NAME=Tech Brief Daily
MAX_ARTICLES=1
MALE_VOICE=en-US-Studio-Q
FEMALE_VOICE=en-US-Studio-O

# ── Logging ──────────────────────────────────────────────────────────────────
LOG_LEVEL=INFO
ENABLE_JSONL_LOGGING=true
ENABLE_S3_LOG_UPLOAD=false

Sources and categories¶

CLIPython API

the-data-packet \
  --sources wired techcrunch \
  --categories security ai \
  --max-articles 2

config = get_config(
    article_sources=["wired", "techcrunch"],
    article_categories=["security", "ai"],
    max_articles_per_source=2,
)

Available sources and categories:

Source	ID	Categories
Wired.com	`wired`	`security`, `science`, `ai`
TechCrunch	`techcrunch`	`ai`, `security`