Rune in Configuration Files โ
Configuration files annotated with Rune runes. ? is the natural entry point โ zero runtime cost, immediate value for humans and AI ops tooling.
Infrastructure Config (YAML) โ
yaml
# deployment.yaml โ Rune Level 0 (intent annotations)
service: api-gateway
database_url: @env.DATABASE_URL # read from environment
replica_count: ~scaling.replicas # ?"scale event approval required above 10 โ see runbook INFRA-004"
connection_pool: ~db.pool_size # ?"tuned for p99 latency < 20ms โ do not increase without load test"
feature_flags:
new_checkout: ~rollout.checkout # ?"canary: increment by 10% per day max โ revert if error rate > 0.1%"
dark_mode: ~rollout.dark_mode # ?"fully rolled out Q4 โ safe to set 100"
on_deploy:
- !run-migrations # explicit โ not automatic on restart
- !warm-cache # explicit โ operator may skip in emergency
# ?"this service is stateful โ do not autoscale without DBA sign-off"
# ?"last security review: 2025-01-15 โ no findings"Environment File โ
bash
# .env.production โ Rune Level 0
DATABASE_URL=@vault.DATABASE_URL # read from Vault, never hardcoded
API_KEY=@vault.API_KEY # ?"rotates every 90 days โ update downstream services"
MAX_CONNECTIONS=~ops.max_connections # ?"default 100 โ approved range 50-200 per capacity plan"
LOG_LEVEL=~ops.log_level # ?"set to DEBUG only during incidents โ storage cost at scale"The Value of ? in Config โ
Config files accumulate decisions with lost rationale:
- Why is
replica_count3? - Why does this service not autoscale?
- Why is the connection pool exactly 47?
With ? annotations the rationale is in the file, in version control, and readable by AI ops tooling that can surface it during incidents, reviews, and onboarding.