config.schema.json says phase_diff is advisory at shipped; route table says off

john commented

2026-08-15 14:33:37 +00:00

Owner

Source: strict re-review C-10
Verdict: AGREE - medium

Claim as reviewed

The phase_diff trigger's schema prose describes shipped behaviour that disagrees with the route table used by actual routing.

At HEAD `f354864` (v3.3.0)

cadence-core/config.schema.json:80 - "Leave it UNSET and the stakes level decides - off at solo, advisory at shipped, adjudicated at critical".

cadence-core/route-table.json shipped row - "phase_diff": "off".

Only the shipped word is wrong; solo (off) and critical (adjudicated) match.

Note

Same root cause as the config.mjs get issue: the schema is stating a default that routing does not honour. A prose assertion comparing every trigger's schema sentence to the table row would close both.

**Source:** strict re-review C-10 **Verdict:** AGREE - medium ### Claim as reviewed The `phase_diff` trigger's schema prose describes shipped behaviour that disagrees with the route table used by actual routing. ### At HEAD f354864 (v3.3.0) `cadence-core/config.schema.json:80` - "Leave it UNSET and the stakes level decides - off at solo, advisory at shipped, adjudicated at critical". `cadence-core/route-table.json` shipped row - `"phase_diff": "off"`. Only the shipped word is wrong; `solo` (`off`) and `critical` (`adjudicated`) match. ### Note Same root cause as the `config.mjs get` issue: the schema is stating a default that routing does not honour. A prose assertion comparing every trigger's schema sentence to the table row would close both.

john added the

documentation

external-review

labels

2026-08-15 14:33:37 +00:00

john added this to the v3.4.1 milestone

2026-08-15 15:01:08 +00:00

john referenced this issue

2026-08-15 15:02:21 +00:00

self-verify never compares a trigger's schema default or prose to route-table.json #135

john commented

2026-08-15 15:02:22 +00:00

Author

Owner

Sequencing: #135 lands first in v3.4.1. Its assertion fails on the current tree, which proves this issue is real before it is touched, and its failure output enumerates the exact trigger rows to fix here. Do not plan this in parallel with #135 - this consumes its output. Full reasoning in the comment on #135.