docs(model-config): remove unbacked 5x quota claim, Sonnet 5 default effort medium (#440) - #443
Merged
Merged
Conversation
…effort medium (#440) The '~5x less Max quota' figure (8 instances, 6 live-guidance files) was traced to commit ab6fc9c with zero supporting measurement. Replaced with qualified language: tokenizer note primary-sourced to Anthropic pricing docs, community cost reports attributed as such, explicit 'no controlled measurement exists — check /usage'. Sonnet 5 documented default effort: high -> medium (CodeRabbit — the source already cited for the max-doubles-cost claim — recommends medium; fresh Fable + Codex xhigh opinions independently converged 2026-07-12). Ladder is now medium -> high -> xhigh across all lanes, tables, skills, and the setup skill's installed settings.json block. hooks/model-effort-check.sh: medium joins the silent floor (the hook must not nag users following the wizard's own default); explicit low and the settings-only-max quirk still warn; warning compacted to 482 bytes worst-case (was 556, over the 500 cap). Tests: 4 new doc-consistency tests (mutation-verified both directions), repo-wide guards for the unqualified-multiplier claim class and the stale high-ladder shape, hook cap test now probes both warning paths and hard-fails on vacuous zero output. Review: Fable round 1 (9 findings, incl. P1 setup-skill config block still installing effortLevel high) + Codex xhigh round 1 (3 P1s, incl. the cap regression) — all fixed, round 2 CERTIFIED.
BaseInfinity
added a commit
that referenced
this pull request
Jul 15, 2026
…net 5 medium default (#448) Catch-up release: seven consumer-affecting commits were stranded on main since v1.86.0 (July 5) because nothing prompts the manual tag-to-publish step — most urgently #445, the Copilot CLI argument-hint fix from the repo's first external contributor (@thejesh23), which npm consumers were not receiving. - CHANGELOG [1.87.0]: leads with the first-external-contribution credit; covers #441 (GPT-5.6 Sol reviewer sweep), #443 (Sonnet 5 medium default, unbacked 5x quota claim removed, hook floor), #445/#447 (Copilot fix + regression test), #446 (Setup A semantics), #440/#442 (chores) - Version bumps at all sites found by the broad grep (v1.85.0 lesson) - ROADMAP #448 (Copilot CLI compat watch) + #449 (release-drift gate — this incident's post-mortem; scope: own-package npm-vs-main watcher) - skills/update: [1.87.0] example line; two oldest bullets trimmed to stay under the 20K session-load cap Cross-model release review: Codex xhigh (GPT-5.6 Sol), 3 rounds to CERTIFIED — caught a sed-falsified historical version citation and two unsupported duration claims in my own incident narrative. All suites green.
BaseInfinity
added a commit
that referenced
this pull request
Aug 12, 2026
…net 5 medium default Catch-up release: seven consumer-affecting commits were stranded on main since v1.86.0 (July 5) because nothing prompts the manual tag-to-publish step — most urgently #445, the Copilot CLI argument-hint fix from the repo's first external contributor (@thejesh23), which npm consumers were not receiving. - CHANGELOG [1.87.0]: leads with the first-external-contribution credit; covers #441 (GPT-5.6 Sol reviewer sweep), #443 (Sonnet 5 medium default, unbacked 5x quota claim removed, hook floor), #445/#447 (Copilot fix + regression test), #446 (Setup A semantics), #440/#442 (chores) - Version bumps at all sites found by the broad grep (v1.85.0 lesson) - ROADMAP #448 (Copilot CLI compat watch) + #449 (release-drift gate — this incident's post-mortem; scope: own-package npm-vs-main watcher) - skills/update: [1.87.0] example line; two oldest bullets trimmed to stay under the 20K session-load cap Cross-model release review: Codex xhigh (GPT-5.6 Sol), 3 rounds to CERTIFIED — caught a sed-falsified historical version citation and two unsupported duration claims in my own incident narrative. All suites green.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
ROADMAP #440
Two coupled corrections to live guidance every consumer repo inherits, plus the hook to match.
1. Unbacked quota claim removed
The "~5x less Max quota" figure (8 instances across 6 files) was traced via git history to
ab6fc9c— no measurement in the commit, CHANGELOG, ROADMAP, or any review artifact. It also survived as rewordings ("fraction of the quota cost", "far less quota burn") which are removed too. Replacement language cites only what's sourced: Anthropic's tokenizer note (primary), community cost reports attributed as such, and an explicit "no controlled Sonnet-5-vs-Opus-4.6 measurement exists — check/usage".2. Sonnet 5 default effort:
high→mediumCodeRabbit — the exact source the wizard already cites for "max doubles cost for marginal gains" — actually recommends medium as the sweet spot. Fresh Fable (85% confidence, reversing its own July 4 position) and Codex xhigh opinions independently converged on medium on 2026-07-12. Ladder is now
medium→high→xhigheverywhere, including the setup skill's installedsettings.jsonblock (Fable's P1: it was still writingeffortLevel: highinto consumer repos).3. Hook floor matches the docs
model-effort-check.sh:mediumjoins the silent set — the hook must not nag users following the wizard's own documented default. Explicitlowand the settings-only-maxquirk still warn. Warning compacted to 482 bytes worst-case (Codex found it at 556, over the 500 cap, hidden by a cap test that had gone vacuous).Verification