From 0b1e4e2a17aa5cb6e4f8c14cdf2c4c3dec79dd3c Mon Sep 17 00:00:00 2001 From: Jeroen Schweitzer Date: Sun, 29 Mar 2026 17:42:13 +0200 Subject: [PATCH] docs(sprints): mark Track C (East Reach) done in copy briefing Co-Authored-By: Claude Opus 4.6 (1M context) --- docs/sprints/sprint-29/copy.md | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/docs/sprints/sprint-29/copy.md b/docs/sprints/sprint-29/copy.md index 67327b0e4..d33baa992 100644 --- a/docs/sprints/sprint-29/copy.md +++ b/docs/sprints/sprint-29/copy.md @@ -18,6 +18,18 @@ Each system follows the `/atlas author-system` pipeline: See `.claude/skills/atlas/references/author-system.md` for full details. +**STRICT: One system at a time.** Complete the full pipeline (steps 1–7) +for a single system before starting the next. Do NOT batch systems for +writing, naming, or review. Batching causes agent context overflow, +cross-system bleed in review feedback, and review agents getting stuck +on oversized payloads. Each system must be isolated in the review agents' +context windows — send one system to Miri, get names back and apply them, +send to Paula and Ozzie, get reviews back and fix issues, verify, commit, +then move to the next system. Between systems, clear reviewer agent +contexts (shutdown + respawn, or user-initiated context clear) to prevent +accumulated review data from prior systems causing bleed or context +pressure in the next review cycle. + ## Work Order Process by corridor, hop-order within each corridor (closest to Gateway first). @@ -27,7 +39,7 @@ All tracks run in series on the `copy` branch (parallel worktree approach supers |-------|-------|----------|---------|-----------|--------| | A | #739 | North Reach | 43 | 4-12 | done | | B | #740 | South Reach | 47 | 4-10 | done | -| C | #741 | East Reach | 47 | 4-10 | hop 4 done (6/47) | +| C | #741 | East Reach | 47 | 4-9 | done | | D | #742 | West Reach | 48 | 4-11 | | | E | #743 | Deep Frontier | 73 | 5-12 | |