🐙
$ git clone https://github.com/Hmbown/CodeWhale.git
; the wedge that took out the owner's own 0.9.9 session).
A concrete route/offering output limit now outranks the conservative
8,192-token compatibility guess for an uncatalogued model. Routes that
publish no output limit remain fail-closed, documented model ceilings stay
authoritative, and a route limit can never raise the requested cap (, ): an Anthropic-family model the catalog does
not describe keeps the 64K Messages floor as its clamp and the ChatGPT/
Codex OAuth route keeps its 4K policy, but
label them
values are unchanged.
Telemetry default-on is visible () is now honored by both
the posture gate and the execution envelope, so
of being refused as "Executes" (, ).
Responses-dialect route (
Anthropic-Messages routes are carried through pi-ai
or refused; only credentialed base URLs are still refused, with an error
that names provider and model (; harvested from ).
macOS CI: the
worker that had already completed and raced the rail's focus; it now holds
the child's wrap-up and waits for a settled live row (refs ).
DeepSeek V4 pricing follows the published peak/off-peak tiers (peak
01:00–04:00 and 06:00–10:00 UTC; off-peak is half of peak) for
each turn's recorded time; the stale single-tier rows understated cost up
to ~4×. Because every direct DeepSeek first-party rate is now
time-windowed, the scorecard fails closed (
undated DeepSeek turn instead of guessing a tier ( follow-up,
verified against api-docs.deepseek.com on 2026-08-17).
SSE UTF-8 split across HTTP/2 DATA frames now fails closed in every
streaming dialect: a shared strict decoder, tail flush, and
approximations, with byte-chunk decoder tests ().
CI:
answered Fleet children with SSE while they call the blocking JSON path;
the parse failure was retried and double-counted the worker on slow macOS
runners ().
Context: every web tool surface (
compacted like shell output instead of consuming the ordinary hard limit
().
Routing: a lowercase saved selector such as
owning Z.ai / DeepSeek catalog row (case-fold fallback, only when exactly
one provider-owned wire id matches) instead of being classified as another
provider's bare model (; diagnosis by ).
Model catalog brought current as of 2026-08-17 against the official
pricing pages: gpt-5.6-terra / gpt-5.6-luna rates,
$2/$10 (the announced September increase was withdrawn),
added,
Mistral first-party rows, xAI
tiers, Gemini and Qwen limits, and RedNote's
OpenRouter row (no first-party API exists yet) — every number carries its
source and a pinned test ().
CI: the release workflows no longer restore npm/cargo caches after
checking out a caller-supplied SHA — the CodeQL cache-poisoning Highs
are closed with a contract test over the workflow files ().
Ghostty's truecolor underwater field now uses a dedicated synchronized
60 FPS lane instead of the legacy 30 FPS compatibility cap, with continuous
caustic fades replacing visibly stepped color changes.
Live reasoning's advertised
composer's first-character paste-burst hold, while spaces in an active paste
remain payload. The newest reasoning preview also spends only genuinely free
viewport rows before truncating instead of stopping at the fixed 10/12-row
fallback on roomy terminals.
Strict
the remaining links are explicit Markdown targets (, refs /).
Configured skill prompts are stable across session roots and operating
systems: only custom configured roots hide their physical path, ordinary
workspace/global skills keep a discoverable privacy-safe path, warning
replacements are boundary-aware (including non-UTF-8 Unix paths), and
Windows separators render as
leaner without raising a runtime-contract ceiling ().
Auto-router classifier requests accept
preserving the existing default when the key is absent ().
The docs shell and shared web components now route localized copy through
the typed dictionary spine; these are two incremental phases of , ).
Documentation: stale A/B/C-tier references, provider defaults, module
descriptions, and line anchors now match the current code ().
Localization: README translations for Français, Deutsch, 繁體中文, हिन्दी,
Türkçe, Italiano, Polski, العربية and Català join the existing nine
().
Docs: README Integrations section (incl. the DeepSeek Harness
path, docs/INTEGRATIONS_DSH.md) localized across all READMEs; RFC keeping
the deterministic-first auto-review hybrid ().
profile via dsh's documented
the 0.9.8 exported-CSS skin that dsh's inline body variables overrode
(docs/design/DSH_BUNDLE_SKIN.md, docs/INTEGRATIONS_DSH.md) ().
Fleet: agent shadowing is visible — a roster-row badge, a Layers block in
agent detail, and a
and
Tests:
rule and maps the Auto-Review guardian acceptance items to the engine
journeys that exercise them ().
Dependencies: ratatui 0.30.2, thiserror 2.0.20.
8,192-token compatibility guess for an uncatalogued model. Routes that
publish no output limit remain fail-closed, documented model ceilings stay
authoritative, and a route limit can never raise the requested cap (, ): an Anthropic-family model the catalog does
not describe keeps the 64K Messages floor as its clamp and the ChatGPT/
Codex OAuth route keeps its 4K policy, but
OutputCeilingSource gained anunverified rung for both, so exec-stream receipts and the model pickerlabel them
unverified/"assumed floor" instead of documented. Clampvalues are unchanged.
the posture gate and the execution envelope, so
git log, find | head,npm view and the other bounded read-only commands run in-place insteadof being refused as "Executes" (, ).
integrations dsh plan no longer refuses DeepSeek's defaultResponses-dialect route (
deepseek-v4-flash); Responses andAnthropic-Messages routes are carried through pi-ai
openai-responses / anthropic-messages instead of being approximatedor refused; only credentialed base URLs are still refused, with an error
that names provider and model (; harvested from ).
agent_focus_pty auto-review receipt test waited on aworker that had already completed and raced the rail's focus; it now holds
the child's wrap-up and waits for a settled live row (refs ).
01:00–04:00 and 06:00–10:00 UTC; off-peak is half of peak) for
deepseek-v4-flash and deepseek-v4-pro in USD and CNY, resolved fromeach turn's recorded time; the stale single-tier rows understated cost up
to ~4×. Because every direct DeepSeek first-party rate is now
time-windowed, the scorecard fails closed (
missing_recorded_time) on anundated DeepSeek turn instead of guessing a tier ( follow-up,
verified against api-docs.deepseek.com on 2026-08-17).
streaming dialect: a shared strict decoder, tail flush, and
decode_failed propagation (InvalidSseUtf8) replace the per-dialectapproximations, with byte-chunk decoder tests ().
release_four_read_only_fleet_roles_launch_with_canonical_promptsanswered Fleet children with SSE while they call the blocking JSON path;
the parse failure was retried and double-counted the worker on slow macOS
runners ().
Web, web_search, web.run,fetch_url) now uses the noisy-result soft limit, so large fetches arecompacted like shell output instead of consuming the ordinary hard limit
().
glm-5.2 resolves against theowning Z.ai / DeepSeek catalog row (case-fold fallback, only when exactly
one provider-owned wire id matches) instead of being classified as another
provider's bare model (; diagnosis by ).
pricing pages: gpt-5.6-terra / gpt-5.6-luna rates,
claude-sonnet-5 keeps$2/$10 (the announced September increase was withdrawn),
claude-opus-5added,
kimi-k3 and kimi-k2.7-code-highspeed, MiniMax-M2.7-highspeed,Mistral first-party rows, xAI
grok-4.5 / grok-4.3 with long-contexttiers, Gemini and Qwen limits, and RedNote's
dots3-note preview as anOpenRouter row (no first-party API exists yet) — every number carries its
source and a pinned test ().
checking out a caller-supplied SHA — the CodeQL cache-poisoning Highs
are closed with a contract test over the workflow files ().
60 FPS lane instead of the legacy 30 FPS compatibility cap, with continuous
caustic fades replacing visibly stepped color changes.
Space:expand action now runs before thecomposer's first-character paste-burst hold, while spaces in an active paste
remain payload. The newest reasoning preview also spends only genuinely free
viewport rows before truncating instead of stopping at the fixed 10/12-row
fallback on roomy terminals.
cargo doc builds no longer fail on bare URLs in rustdoc comments;the remaining links are explicit Markdown targets (, refs /).
systems: only custom configured roots hide their physical path, ordinary
workspace/global skills keep a discoverable privacy-safe path, warning
replacements are boundary-aware (including non-UTF-8 Unix paths), and
Windows separators render as
/. The skills prompt is also 50 bytesleaner without raising a runtime-contract ceiling ().
[auto.router] timeout_secs, whilepreserving the existing default when the key is absent ().
the typed dictionary spine; these are two incremental phases of , ).
descriptions, and line anchors now match the current code ().
Türkçe, Italiano, Polski, العربية and Català join the existing nine
().
dsh pluginpath, docs/INTEGRATIONS_DSH.md) localized across all READMEs; RFC keeping
the deterministic-first auto-review hybrid ().
dsh integration: the Codewhale palette is applied through the bundleprofile via dsh's documented
overrideTokens (on by default;codewhale integrations dsh update --skin false turns it off), replacingthe 0.9.8 exported-CSS skin that dsh's inline body variables overrode
(docs/design/DSH_BUNDLE_SKIN.md, docs/INTEGRATIONS_DSH.md) ().
agent detail, and a
doctor "Fleet roster layers" section (JSONoperate_fleet.roster.multi_layer), in all 15 TUI locales. Layer collapseand
[fleet.profiles] migration stay for 0.9.10 ().crates/tui/tests/README.md states the keyless assembled-journeyrule and maps the Auto-Review guardian acceptance items to the engine
journeys that exercise them ().
Removed
dshintegration: the exported-CSS skin file and its "skin export" status
line (superseded by the bundle-appliedoverrideTokensskin, ) — a concrete route/offering output limit outranks the
8,192-token compatibility guess for an uncatalogued model (); web tool results use the noisy soft limit (); and configured-skill prompts stay
stable across ephemeral roots and operating systems ().- Gabriel-Degret (; first contribution).
- (carried upstream in
(context-
window honesty, (bwrap extra roots,
(SSE UTF-8 garbling on DeepSeek Flash, (unverified live
pricing, (the macOS filtered-suite hang
investigation on for full notes and docs/CHANGELOG_ARCHIVE.md for older releases.
Vollständiger Original-Bericht
Ausführliche Details, Code-Beispiele & Hersteller-Stellungnahme auf github.com.
Wie bewertest du diesen Beitrag?
1 Klick Feedback Teilen mit Netzwerk & Team:
Hat Ihnen dieser Tipp / Anleitung geholfen?
Community-Analysen & Experten-Meinungen 0
Verfasse deine eigene Analyse, teile Workarounds oder diskutiere diesen Vorfall im Blog.
Noch keine Community-Analyse verfasst. Markiere einen Textabschnitt oder klicke oben auf „ Eigene Analyse verfassen“!
Community Pulse: Relevanz-Einschätzung
1 Klick Experten-Votum 🔴 Akute Relevanz 52%
🟡 In Evaluierung 29%
🟢 Keine Auswirkung 14%
Spannende Innovation 5%
Verwandte Story-Cluster & Quellen (Vektor-KI)
1 Quelle
Generate playable HTML5 games with Mistral Medium and GLM
1 Quelle
New AI waifus, new Deepseek, realtime worlds, Happy Shrimp, tiny TTS: AI NEWS
1 Quelle
Ox Alpha reveal, realtime Minimax, Qwen Next, Hy4, robot olympics: AI NEWS
Tipp: Mit Pfeiltasten [ ← ] und [ → ] blättern
SOCIAL SHARE CARD GENERATOR