PLAN rev 3: TOML config

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

PLAN.md

@@ -1,6 +1,6 @@
 # crossbar — an affinity router for the fleet's llama-servers
 
-**Status:** plan, 2026-09-25 (rev 2: SQLite state + accounting, Kyle). Nothing built. Go.
+**Status:** plan, 2026-09-25 (rev 2: SQLite state + accounting; rev 3: TOML config — Kyle). Nothing built. Go.
 **Owner:** Kyle. Drafted by claude from the 2026-09-25 discussion (Inference Infrastructure).
 
 ## 1. Problem
@@ -145,35 +145,47 @@ exposes the rollups; a nightly job prunes `requests` older than `retention`
 
 Body contents are never stored — only counts and timings.
 
-## 8. Config (`crossbar.yaml`)
+## 8. Config (`crossbar.toml`)
 
-```yaml
+TOML (`github.com/BurntSushi/toml`): no indentation semantics, no implicit
-listen: "100.x.y.z:7777"        # tailnet address only; never 0.0.0.0
+type coercion, and it is what Kyle's Rust projects already use. A NixOS module
-db: /var/lib/crossbar/crossbar.db
+can generate it with `pkgs.formats.toml`.
-retention: 180d
+
-poll_interval: 60s
+```toml
-lease_idle: 30m
+listen        = "100.x.y.z:7777"   # tailnet address only; never 0.0.0.0
-queue_max: 8
+db            = "/var/lib/crossbar/crossbar.db"
-hosts:
+poll_interval = "60s"
-  straylight:
+lease_idle    = "30m"
-    base_url: http://straylight.<tailnet>:11434
+queue_max     = 8
-    weight: 1.0
+retention     = "180d"
-    models: { ornith-1.5-35b-a3b: {parallel: 4}, ornith-1.5-9b-uncensored: {parallel: 6} }
+
-  titan:
+[hosts.straylight]
-    base_url: http://titan.<tailnet>:8081
+base_url = "http://straylight.<tailnet>:11434"
-    weight: 2.0                 # ~2× straylight decode
+weight   = 1.0
-    models: { ornith-1.5-35b-a3b: {parallel: 4}, laguna-s-2.1: {parallel: 2} }
+models   = { "ornith-1.5-35b-a3b" = { parallel = 4 }, "ornith-1.5-9b-uncensored" = { parallel = 6 } }
-  dixie:
+
-    base_url: http://dixie.<tailnet>:11434
+[hosts.titan]
-    weight: 0.8
+base_url = "http://titan.<tailnet>:8081"
-    models: { ornith-1.5-9b-uncensored: {parallel: 6} }
+weight   = 2.0                       # ~2x straylight decode
-routes:                          # optional defaults / pins
+models   = { "ornith-1.5-35b-a3b" = { parallel = 4 }, "laguna-s-2.1" = { parallel = 2 } }
-  opencode-a: { default_model: ornith-1.5-35b-a3b }
+
-  opencode-b: { default_model: ornith-1.5-35b-a3b }
+[hosts.dixie]
-  paper:      { default_model: qwen3.8-27b-uncensored, pin: titan }
+base_url = "http://dixie.<tailnet>:11434"
-  hermes-straylight: {}
+weight   = 0.8
-  hermes-talos: {}
+models   = { "ornith-1.5-9b-uncensored" = { parallel = 6 } }
-  hermes-titan: { pin: titan }   # titan's models are the titan agent's first
+
+# optional per-route defaults / pins
+[routes.opencode-a]
+default_model = "ornith-1.5-35b-a3b"
+[routes.opencode-b]
+default_model = "ornith-1.5-35b-a3b"
+[routes.paper]
+default_model = "qwen3.8-27b-uncensored"
+pin = "titan"
+[routes.hermes-straylight]
+[routes.hermes-talos]
+[routes.hermes-titan]
+pin = "titan"                        # titan's models are the titan agent's first
 ```
 
 Client side, no code changes:
@@ -201,7 +213,7 @@ Client side, no code changes:
 
 ## 10. Milestones
 
-- **v0 (a day):** config, static routes → host mapping, reverse proxy with
+- **v0 (a day):** TOML config, static routes → host mapping, reverse proxy with
   streaming, `/health` + `/models` poller, passive health, `/_crossbar/hosts`.
   Replaces the hand-maintained provider lists. No leases yet: each route has
   a fixed host list in preference order; first healthy wins.