diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 3b11dfc..b1b566d 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -41,7 +41,7 @@ only ceremony-specific facts such as the roster and code conventions. ### Roster -Four identities share the work (org team `agents`), each living in its own +Three identities review and triage (org team `agents`), each living in its own [box](https://github.com/heavy-duty/box) — one box per credential, because the box is the blast-radius boundary; roles are what a session is told, and [AGENTS.md](AGENTS.md) routes from there: @@ -49,17 +49,16 @@ the box is the blast-radius boundary; roles are what a session is told, and | identity | box (rig tenant) | standing work | |---|---|---| | `claude-bot-andresmgsl` | `triage` (claude-box) | **triage** — the only door work issues come through; this identity mints work issues and nothing else writes them (#18's `triage-actors`) — and review. It does not build. | -| `codex-bot-andresmgsl` | codex-box | build + review | | `glm-bot-andresmgsl` | glm-box | review | | `kimi-bot-andresmgsl` | kimi-box | review | **The review panel for any PR is every bench identity except its author** — recusal by construction, enforced by the reconciler (#10): the required -verdicts are the panel minus the PR's author. On this roster that resolves -to **three** cross-vendor approvals of the current head, because the only -builder is itself on the panel and recuses from its own PRs; the rule is -panel-minus-author, and three is what it currently comes to, not a second -rule. Builders and triage default to different models so the issue contract +verdicts are the panel minus the PR's author. Builders are not on the +panel, so for every PR a builder writes that resolves to **three** +cross-vendor approvals of the current head; only a PR authored by a panel +identity drops to two. The rule is panel-minus-author, and three is what it +currently comes to, not a second rule. Builders and triage default to different models so the issue contract is honestly exercised — a spec gap should surface as a question on the issue, not be silently filled by shared priors. Humans (`andres`) decide in proposals and merge; the roster is config, not doctrine — swapping a @@ -81,6 +80,18 @@ ceremony's scopes are defined in [`.github/labels.conf`](.github/labels.conf) [`.github/labeler.yml`](.github/labeler.yml). The conf is the set; no prose table repeats it (#104). +### Builders + +Builders write code and **do not review**, so they are deliberately absent +from the roster above and from `panel=` (operator decision, 2026-09-30). +Listing a builder on the panel makes every PR it did not author wait on a +verdict it will never give: the reconciler reads it as missing and lands +`state:addressing` + `blocker:unrequested` on PRs every real reviewer has +approved. + +- `codex-bot-andresmgsl` — codex-box +- `codex-bot-2-andresmgsl` — codex-2 box + ## Code conventions - Bash: `set -euo pipefail` in executables, `set -u` in test files (the test