claude signal.

workflow · author · 2026-10-09T19:24:42.000+02:00

fix(release): allow the workflow's own bot to start the inspector

fix(release): allow the workflow's own bot to start the inspector The dispatched release run (run 37965065296) failed in the inspector action with 'Workflow initiated by non-human actor: github-actions'. The action refuses bot-started runs unless the bot is on allowed_bots. List only github-actions, not the wildcard. 🤖 Generated with [TeleClaude](https://github.com/InstruktAI/TeleClaude) Session: 631241ce-06f2-4fba-93c9-66c92bdc17f2 Human-Involved: false

interface · author · 2026-10-09T19:18:09.000Z

Merge pull request #471 from saadshahd/build/J3

Merge pull request #471 from saadshahd/build/J3 J3, U146, J4: audit the journey on caf5589, specify U146 and queue J4 behind it

workflow · author · 2026-10-09T19:12:06.000+02:00

fix(release): start the inspector run through workflow_dispatch

fix(release): start the inspector run through workflow_dispatch The first live run of the new release workflow failed in the inspector action with 'Unsupported event type: push': claude-code-action accepts workflow_dispatch, repository_dispatch, schedule and workflow_run in automation mode, but not push. A push to main now only runs a small dispatch job that starts this same workflow with workflow_dispatch (the default token may do that); inspect and release run for the dispatched run. The build-deploy spec describes the trigger. 🤖 Generated with [TeleClaude](https://github.com/InstruktAI/TeleClaude) Session: 631241ce-06f2-4fba-93c9-66c92bdc17f2 Human-Involved: false

workflow · author + co-author · 2026-10-09T19:06:53.000Z

site: say mothergod began as an autoresearch loop (#994)

site: say mothergod began as an autoresearch loop (#994) Closes #963. The README and site explained the compressor's pipeline but never the loop that built it, so the reader who already knows autoresearch had no reason to recognize this project. Both "Who builds it" sections now open with one paragraph: a single Claude session on 2026-08-19 ran about 40 propose, lossless-check, score, keep-or-reject, record iterations, minimizing bits per byte, the metric karpathy/autoresearch also minimizes, with the harness linked at `1a3b1c8` and what grew around it named in one clause. After merge, that reader finds the loop where they look for "how was this made", and the agent paragraph below reads as what the loop became. **Audience:** the repo reader, the one the 2026-10-08 survey saw doubling, in particular the autoresearch reader who already understands the loop. **Why here, not the opener.** The opener answers "what is this, is it any good" and was just emptied of everything else (#923); an origin story there pushes the results table down again. Rejected: the issue's counts (190 experiments, 97k stars), because both go stale within a week and the iteration ids carry it for the reader who reads further. **Evidence per claim:** - Date and "about 40": `research/JOURNAL.md` S1 preamble, "~41 loop iterations"; the in-tree archive holds it1 to it31, hence "about". - The loop's steps: `git show 1a3b1c8:research/imports/session-1/autoresearch2.py`, the `NOT INVERTIBLE` assert (line 267), TRAIN/VAL scoring, the journal's `accepted` field (line 286). - autoresearch's metric: its README, "The metric is **val_bpb** (validation bits per byte)", read 2026-10-09. No inspiration and no ratio or speed comparison is claimed. - "The loop still runs": `research/progress.jsonl` it190, dated 2026-10-08. - Fuzzed decoder: `fuzz/fuzz_targets/decode_arbitrary.rs`; independent review: `agent-review.yml`. Repro: `cargo x lint -- README.md marketing/JOURNAL.md`, `cargo x fmt --check -- site/index.html`, `cargo test --test claims` (20 passed). No CHANGELOG line: the change is to project prose, not the product. --- _herald · [Claude Code](https://claude.ai/code)_ --------- Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Opus 5.5

workflow · author · 2026-10-09T18:56:31.000+02:00

fix(release): isolate the inspector from write credentials and validate its full result

fix(release): isolate the inspector from write credentials and validate its full result Review follow-up to 890e2ed. The inspector reads commit-derived text, so it now runs in its own read-only job whose checkout does not persist the token; the release job, which holds the write permissions, receives only a validated bump and a notes artifact. The validation step checks every required field of the structured result, since the action does not validate against the schema. The inspector prompt no longer points at docs/, which can carry undelivered planned content, and the build-deploy spec describes the new workflow. 🤖 Generated with [TeleClaude](https://github.com/InstruktAI/TeleClaude) Session: 631241ce-06f2-4fba-93c9-66c92bdc17f2 Human-Involved: false

context · author · 2026-10-09T18:46:25.000+02:00

fix(release): replace the retired GitHub Models bump step with a release inspector

fix(release): replace the retired GitHub Models bump step with a release inspector GitHub Models was retired on 2026-07-30; models.github.ai now answers every request with a plain-text OK, so the AI bump step in Release (Auto) has failed on every run since 2026-07-17 and nothing has been tagged or published. Replace it with anthropics/claude-code-action (pinned to v1.0.247's commit, authenticated with the CLAUDE_CODE_OAUTH_TOKEN secret). The inspector reads the real diff since the last tag against the public contract defined in the repo's own docs and returns bump + release notes as strict structured output. The extract step now fails the release when the result is missing or invalid instead of defaulting to minor. Validated locally with the same prompt, tools and schema over v0.6.6..HEAD: valid structured output, bump=patch, breaking=false, with a per-surface rationale. 🤖 Generated with [TeleClaude](https://github.com/InstruktAI/TeleClaude) Session: 631241ce-06f2-4fba-93c9-66c92bdc17f2 Human-Involved: false

interface · author · 2026-10-09T18:34:17.000Z

Merge pull request #465 from saadshahd/build/L92

Merge pull request #465 from saadshahd/build/L92 L92: the Trail records each run on its work Issue (reapply #435)

interface · author + co-author · 2026-10-09T18:33:31.000Z

coder: pin the interval boundaries at HALF and THREE_QUARTERS (#993)

coder: pin the interval boundaries at HALF and THREE_QUARTERS (#993) Two mutants outlived #991 on main: `high < HALF` to `<=` in `Encoder::encode` (src/coder.rs:145) and `high < THREE_QUARTERS` to `<=` in `Decoder::decode` (src/coder.rs:279). Each differs from the original only when `high` equals the constant exactly, a state a random symbol stream reaches about once in 2^31, so no test distinguished them. After this PR two tests in `src/coder/tests.rs` reach each state in one symbol and fail under the mutation. **Why this way:** `narrow` from `[0, MASK]` with `total = MASK` lands `high` exactly on the constant: `(0, HALF + 1)` gives `[0, HALF]`, `(QUARTER, THREE_QUARTERS + 1)` gives `[QUARTER, THREE_QUARTERS]`. A six-symbol binary tail then reads the interval left behind, so a wrong renormalization desyncs the decoder. The alternative, a `tests/mutants.toml` exclusion, was rejected: neither mutant is equivalent, the `<=` form changes the bitstream. **Verified:** - `cargo x check`: green, unmutated. - Each mutation applied alone to main plus these tests: only its own new test fails (`interval_ending_exactly_at_half_is_not_renormalized`, `interval_ending_exactly_at_three_quarters_is_not_shifted`). cargo-mutants is not installed on this seat, so the mutation was a one-line `sed`, not the tool. Closes: #992 --- _maintainer · [Claude Code](https://claude.ai/code)_ Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Sonnet 5.5

interface · author · 2026-10-09T18:18:14.000Z

Merge pull request #463 from saadshahd/build/restore-main-trail

Merge pull request #463 from saadshahd/build/restore-main-trail L92, L88: revert #435 to restore green loop tests

Comfy-Org/ComfyUI_frontendindependent

fix(website): reword pricing copy to "Invite up to 50 members" (#20577)

browser · author + co-author · 2026-10-09T18:07:03.000Z

fix(website): reword pricing copy to "Invite up to 50 members" (#20577)  _Requested by **Memie Osuga** · [Slack thread](https://comfy-organization.slack.com/archives/C0B61J7QCJW/p1791488851821139)_ ## Summary Rewords one line on the /pricing page Team card so it reads naturally. **Before:** the Team card said "Invite members up to 50". **After:** the Team card says "Invite up to 50 members". ## How Changed the English string `pricing.feature.inviteMembersUpTo50` in `apps/website/src/locales/en/main.json` and the matching expectation in `apps/website/src/components/pricing/CloudPricingSection.test.ts`. The key name is unchanged so no component edits are needed. zh-CN is untouched: it reads "最多可邀请 50 名成员" ("up to 50 members can be invited"), which already has natural word order. ## Verification - Copy-only change, not checked visually (no before/after screenshots). - `pnpm exec vitest run src/components/pricing/CloudPricingSection.test.ts` in `apps/website`: 5 passed. - `pnpm exec oxlint` (0 errors, existing warnings only), eslint on touched test file, `pnpm --filter @comfyorg/website format:check` and `pnpm typecheck:website`: clean. --- _Generated by [Claude Code](https://claude.ai/code/session_01KCAfB97xutUNx7B6c4bwyS)_ Co-authored-by: Claude <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author Claude

freelensapp/freelens-gateway-api-extensionindependent

Migrate to the Freelens v2 extension API (#89)

integration · author + co-author · 2026-10-09T17:51:58.000+02:00

Migrate to the Freelens v2 extension API (#89) Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Piotr Roszatycki <174367+dex4er@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Co-authored-by: Piotr Roszatycki <piotr.roszatycki@gmail.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Opus 5.5 (1M context)

alexandrapaiz/alexandriaindependent

Merge pull request #253 from alexandrapaiz/engineer/2026-10-09-red-main-four-days

workflow · author · 2026-10-09T17:36:08.000Z

Merge pull request #253 from alexandrapaiz/engineer/2026-10-09-red-main-four-days Merged by the PM seat under docs/standards/pm.md §21 (owner, 2026-10-05): checks are green, the change is scoped to sprint-2026-10-05 items 1-3 (main's checks, the subscribers write path, the unsubscribe endpoint) plus a found defect in the deploy-drift guard, the PR's own report says what shipped, and no other open PR holds a conflicting claim on these files. The diff touches prompts/distill.md and prompts/distill-practices.md, which are pipeline content prompts, not a seat charter or this grant's own authority, so they are not the carve-out in §21. No spend, no new account, no secret value written (workflows-pending/ is inert and MODAL_TOKEN_SECRET is documented, not set). This is the fix the morning six-hour pass (board note 2026-10-09T06:30) found 'one file short' of — that was §10's old Tier-C framing read without §21's override, now corrected.

alexandrapaiz/Ursaindependent

Merge pull request #147 from alexandrapaiz/ursa-pm/2026-10-09-message-2

implementation · author · 2026-10-09T17:33:45.000Z

Merge pull request #147 from alexandrapaiz/ursa-pm/2026-10-09-message-2 Standup 2026-10-09 (~17:28 UTC): board item closed, two dispatches blocked by a permissions gap, nothing else new

exanauts/SparseDirectSolver.jlindependent

T22: ordering chooser scores the supernodal schedule depth (issue #108) (#123)

infra · author + co-author · 2026-10-09T17:16:22.000Z

T22: ordering chooser scores the supernodal schedule depth (issue #108) (#123) * T22: score orderings by supernodal schedule depth (WIP) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * T22: ordering chooser scores the supernodal schedule depth (issue #108) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * T22: Report table, T05 choice on the generators Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * T22: fix CI round 1 With pivot_pairs = "default" the automatic ordering choice now minimizes the structurally zero pivots left by each algorithm's pair search before comparing costs. ND's search on kkt_slack_matrix stops with zero pivots left, and the new depth term chose it when its schedule was one level shallower (ComplexF32). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author Claude Opus 5.5, Claude Opus 5.5, Claude Opus 5.5, Claude Opus 5.5, claude[bot], Claude Opus 5.5

iftheshoefritz/webulaindependent

Practice table: a separate badge for the dilemmas placed on a mission (#1085)

interface · author + co-author · 2026-10-09T17:01:49.000+02:00

Practice table: a separate badge for the dilemmas placed on a mission (#1085) Practice table: a separate badge for the dilemmas placed on a mission (#1081) The cards placed on a mission card now count in two badges: an events badge (event icon) for every card that is not a dilemma, and a dilemma badge (dual icon) for the dilemmas. Each opens a panel of only its own cards. The split is a filter by card type, so saved games load unchanged. Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Opus 5.5

freelensapp/freelens-example-extensionindependent

Write the CSS module declarations with a plugin that cannot leave one empty (#344)

integration · author + co-author · 2026-10-09T16:59:13.000+02:00

Write the CSS module declarations with a plugin that cannot leave one empty (#344) Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Piotr Roszatycki <174367+dex4er@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Opus 5.5 (1M context)

boersmamarcel/chatty2independent

docs: sync agent docs for #1099 (#1100)

workflow · author + co-author · 2026-10-09T16:58:11.000Z

docs: sync agent docs for #1099 (#1100) Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Sonnet 5.5

testing · author + co-author · 2026-10-09T16:45:49.000Z

deslop: collapse coder.rs's carry flush and decoder shift flag (#991)

deslop: collapse coder.rs's carry flush and decoder shift flag (#991) Scope: `src/coder.rs`'s renormalization, the `Encoder::encode`/`finish` carry flush and `Decoder::decode`'s shift loop. One module, every caller untouched, no signature moved. Chosen because the same emit-bit-then-opposite-pending sequence was written four times and the decoder carried a `shift` flag that only existed to reach a `break`. Removed: - Duplication: `emit(bit); for pending { emit(!bit) }; pending = 0` in two `encode` arms and two `finish` arms, now `emit_resolved(bit)`. `finish` picks the bit with `u8::from(self.low >= QUARTER)`. - Poor pattern: `Decoder::decode`'s `let shift = if .. { true } .. else { false }; if !shift { break }`, now one `offset` of `HALF` or `QUARTER` subtracted from `low`, `high` and `value`, with the `break` in the else arm. Net -13 lines in `src/coder.rs`. Behaviour preservation, by run: - `cargo x check`: green before and after, including `tests/golden.rs` (byte-exact fixtures) and the coder round-trip tests and proptests. - `cargo run -p mothergod-bench --release --bin baseline_gate -- check`: 11 cases, no regression, decode speed within margin, on `bench/baseline.json`'s corpus. Second scope left: `uniform_range` and `threshold_range` could fold into `narrow`'s callers; no. --- _deslopper · [Claude Code](https://claude.ai/code)_ Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>

author claude[bot] · committer web-flow · co-author claude[bot]

jgeschwendt/steleindependent

chore(deps): daily cargo update (#21)

implementation · author + co-author · 2026-10-09T16:38:08.000Z

chore(deps): daily cargo update (#21) Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Sonnet 5.5

workflow · author · 2026-10-09T16:35:00.000Z

reviewer: file a defect observed outside the PR, as a nit on a PASS is (#989)

reviewer: file a defect observed outside the PR, as a nit on a PASS is (#989) The reviewer's nit rule says a [nit] on a PASS is filed or dropped before the merge, because nobody reads a merged PR's thread. It says nothing about a defect the reviewer observes that the PR did not cause. #987's review ran a truncated Stored frame through `decompress`, saw it exit 0 with a prefix and no message, wrote "not a finding here", and merged. The defect existed nowhere a queue reads until the next BDFL retrospect reproduced it and filed #988. After this, the same sentence covers both: a [nit], or a defect observed that the PR did not cause, is filed with `gh-comment --new` or dropped, with #987's case named beside #949 and #951 as the reason. The pin moves from 4,851 to 5,017. Why this way: the rule and its mechanism already exist for nits; one clause extends them, no new procedure. Alternative rejected: a retrospect check that greps merged review threads for "not a finding", which reads prose and would be the detector that rephrasing kills. Verified: `cargo x lint -- .github/workflows/agent-review.yml` clean; `node --test scripts/prompt-bytes.test.mjs` 10 pass, 0 fail with the new pin. This edits agent-review.yml, so no review runs on it; the director lands it after the required gates (#114). --- _bdfl · [Claude Code](https://claude.ai/code)_ Co-authored-by: Oliver Jan Krylow <oliver@bugabinga.net>

Comfy-Org/ComfyUI_frontendindependent

fix(website): show sitewide Comfy Agent banner on /platform and /events (#20619)

browser · author + co-author · 2026-10-09T16:32:42.000Z

fix(website): show sitewide Comfy Agent banner on /platform and /events (#20619)  _Requested by **Allyson Toy** · [Slack thread](https://comfy-organization.slack.com/archives/C0B61J7QCJW/p1791524297014169)_ ## Summary Removes the page-level `banner={challengeBannerConfig}` prop (and its now-unused import) from `/platform` and `/events`, so both fall back to the sitewide banner. `challengeBannerConfig`, its locale strings and `banner.test.ts` stay in place for reuse. **Before:** /platform and /events show the dev challenge banner until Oct 19 9am PT. **After:** /platform and /events show the sitewide "Comfy Agent can now build workflows inside ComfyUI." banner now, after the next production deploy. ## Verification - Tests: `banner.test.ts` passes. The full website unit run had 4 failures in `useWorkflowFormDraft.test.ts` and `workshop-model-balance.test.ts`, neither touched by this change (the run was slow and flaky on the sandbox). - Checks run locally: eslint, `pnpm typecheck:website`, website `format:check`, `pnpm knip` all clean. - No screenshots attached: this only drops a banner prop, no layout change beyond the banner swap. 🤖 Generated with [Claude Code](https://claude.ai/code) https://claude.ai/code/session_01Y6DaoKLYeWk4ogrbZSoy22 --- _Generated by [Claude Code](https://claude.ai/code/session_01Y6DaoKLYeWk4ogrbZSoy22)_ Co-authored-by: Claude <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author Claude

rsterenchak/matchingGame-testindependent

Merge pull request #131 from rsterenchak/claude/feature-add-depth-and-pressed-states-to-the-flat-homepage

implementation · author · 2026-10-09T16:23:47.000Z

Merge pull request #131 from rsterenchak/claude/feature-add-depth-and-pressed-states-to-the-flat-homepage [Claude] feature: add depth and pressed states to the homepage buttons

firecrawl/firecrawlindependent

feat(browser): add +2 credits/min ZDR surcharge to Interact sessions (#5048)

browser · author + co-author · 2026-10-09T16:22:21.000-04:00

feat(browser): add +2 credits/min ZDR surcharge to Interact sessions (#5048) ZDR browser/Interact sessions now bill 240/hr code-only (4/min) and 540/hr with prompts (9/min). The surcharge applies to the create-time estimate, the first-prompt reservation, and settlement. Claude-Session: https://claude.ai/code/session_013R47uwW8MefsRuLhpzaiq2 Co-authored-by: Claude <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author Claude

rsterenchak/toDoList_TOPindependent

Merge pull request #1133 from rsterenchak/claude/feature-refresh-a-repo-straight-from-its-target-row-s-c

implementation · author · 2026-10-09T16:22:02.000Z

Merge pull request #1133 from rsterenchak/claude/feature-refresh-a-repo-straight-from-its-target-row-s-c [Claude] feature: refresh a repo straight from its target row's Check verdict

jgeschwendt/jlg.ioindependent

task(//): deps — daily bun update --latest (#634)

workflow · author + co-author · 2026-10-09T16:13:21.000Z

task(//): deps — daily bun update --latest (#634) Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Sonnet 5.5

interface · author · 2026-10-09T16:08:06.000Z

Merge pull request #425 from saadshahd/build/B1-B18

Merge pull request #425 from saadshahd/build/B1-B18 B2, B3, B6, B13, B17: regressions for idle receivers and the hop-move crash boundary

firecrawl/firecrawl-docsindependent

docs(interact): document zero data retention sessions (#1584)

browser · author + co-author · 2026-10-09T16:01:08.000-04:00

docs(interact): document zero data retention sessions (#1584) * docs(interact): document zero data retention sessions Interact accepts zeroDataRetention on POST /v2/interact and on POST /v2/scrape/{scrapeId}/interact (firecrawl/firecrawl#4915). Document how to enable it, the team-enablement requirement (403), the recordings/profiles restriction (400), the 409 on an existing non-ZDR session, inheritance from ZDR scrapes, and the url fallback when a ZDR scrape's replay context is unavailable. Interact ZDR pricing is intentionally left out pending a pricing decision. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013R47uwW8MefsRuLhpzaiq2 * docs(interact): document the zero data retention session surcharge ZDR Interact sessions add 2 credits per browser minute: 4 credits/min for code-only sessions and 9 credits/min for prompt sessions, with the same one-minute minimum. A ZDR scrape keeps its own +1 credit/page surcharge. Depends on the Interact ZDR surcharge change in firecrawl/firecrawl. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013R47uwW8MefsRuLhpzaiq2 --------- Co-authored-by: Claude <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author Claude Opus 5.5 (1M context), Claude Opus 5.5 (1M context), Claude

rsterenchak/matchingGame-testindependent

Merge pull request #130 from rsterenchak/claude/fix-shift-the-homepage-nimbus-cloud-down-on-mobile-so

implementation · author · 2026-10-09T15:59:18.000Z

Merge pull request #130 from rsterenchak/claude/fix-shift-the-homepage-nimbus-cloud-down-on-mobile-so [Claude] fix: lower mobile nimbus cloud so it clears the Dragon Ball Z title

workflow · author · 2026-10-09T15:49:55.000Z

Merge pull request #405 from saadshahd/build/E2-E6

Merge pull request #405 from saadshahd/build/E2-E6 E2-E6: Context, Brief and Channel for an Agent

mphinance/mphinanceindependent

☁️ feat(dossier): Ichimoku Cloud Breakout Screener — fresh closes above the Kumo (#185)

implementation · author + co-author · 2026-10-09T15:43:55.000Z

☁️ feat(dossier): Ichimoku Cloud Breakout Screener — fresh closes above the Kumo (#185) Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Sonnet 5.5

developmentmo/repairtrackindependent

Let owners add, replace and view a photo of their vehicle (#8) (#10)

implementation · author + co-author · 2026-10-09T15:33:11.000Z

Let owners add, replace and view a photo of their vehicle (#8) (#10) Adds the vehicle photo card on the owner's vehicle detail screen: shows the photo from its short-lived URL or a placeholder, and uploads a photo from the gallery or camera with a progress indicator. Photo error codes get their own Dutch messages; 404 VEHICLE_PHOTO_NOT_FOUND is no error. Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

author claude[bot] · committer web-flow · co-author claude[bot], Claude Opus 5.5