Skip to content

devlog: Rez persona port, and why the roster counts are untested - #34

Merged
d6veteran merged 1 commit into
mainfrom
claude/rez-persona-port-5ovotj
Sep 4, 2026
Merged

devlog: Rez persona port, and why the roster counts are untested#34
d6veteran merged 1 commit into
mainfrom
claude/rez-persona-port-5ovotj

Conversation

@d6veteran

Copy link
Copy Markdown
Collaborator

Two DEVLOG entries covering the work merged in #33. Documentation only; no code changes.

Why these entries

The commit messages in #33 record what changed. They do not record the two decisions most likely to be re-litigated later, and one of those produced no commit at all, so it would otherwise leave no trace.

[2026-09-04] Rez lands as persona eighteen, and every surface a persona touches (milestone)

Records that the profile was already structurally conformant with the other seventeen, verified across twelve canonical heading slots, with generation reproducing the committed artifacts byte for byte. What actually broke was the surrounding surfaces: the TEAM.md section landed under ## The Game Development Team whose intro names a closed set of four, tiers.conf had the new entry outside its tier block where seventeen of eighteen sit inside theirs, and the role name read as a topic in a column where every peer reads as a job.

[2026-09-04] Roster counts removed from the docs, and deliberately not tested (infrastructure)

Records why the docs stopped stating a roster size, and why nothing enforces it. The sweep that first fixed the counts was case sensitive, so its own verification step reported clean while the tagline still read "Seventeen specialists"; two further instances were derived counts that match no search for a roster number; gtm.md had been advertising the wrong team size since before the eighteenth persona existed.

The second entry also records a decision that produced no diff: a regression guard was drafted and abandoned. It would have rejected statements that are true when written, and it required an allowlist for the ASD-STE100 writing-rule counts and the four-person studio. That extends two precedents already in this log, both from 2026-07-31: Robin declining a README/ROADMAP byte-sync test because it would enforce agreement rather than truth, and the banned-claim check deliberately stripping quoted spans rather than maintaining a file exclusion list.

Review notes

No supersession markers. All 27 existing entries were scanned at the heading, summary, and decisions level; nothing is contradicted. The 2026-07-31 doc-assertions entry is extended rather than reversed, since it added assertions that derive both sides from the filesystem while this declines one that could not. "Iris as persona #17" remains accurate as history.

No archiving. The log is at 29 entries and 1037 lines, below the 50-entry and 1500-line thresholds.

Em dashes in the new entries appear only as delimiters in Related list items, matching the existing entries.

bash tests/run.sh passes 296/296.

🤖 Generated with Claude Code

https://claude.ai/code/session_018ZKH7A7XD4eMHrbHKjhKpg


Generated by Claude Code

Two entries.

The first records that Rez's profile was already structurally conformant with
the other seventeen, and that what broke was the surfaces around it: the
TEAM.md section filed under the game studio, a tiers.conf entry outside its
tier block, and a role name that read as a topic rather than a job.

The second records why the docs no longer state a roster count, and why no
test enforces that. The count is not a fact anyone needs, so asserting it was
the defect; removing the assertion removes the failure mode rather than
mitigating it. A guard was drafted and abandoned because it would reject
statements that are true when written, and because it needed an allowlist for
writing-rule counts and the four-person studio. That extends two precedents
already in this log: Robin declining a byte-sync test for enforcing agreement
rather than truth, and the banned-claim check preferring quote-stripping to a
file exclusion list.

No supersession markers: nothing in the existing entries is contradicted.

tests/run.sh: 296/296 passing.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018ZKH7A7XD4eMHrbHKjhKpg
@d6veteran
d6veteran marked this pull request as ready for review September 4, 2026 02:17
@d6veteran
d6veteran merged commit 912ac61 into main Sep 4, 2026
4 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants