# Reported issues for skylight-mcp Pod holds 8 of 8 problems reported by people outside the maintainer team. Issues filed by the project's own owners, members and collaborators are excluded entirely — a maintainer's release checklist is not a warning to a prospective user. Back to [skylight-mcp](/mcp/skylight-mcp). ## Most discussed ### Auto-review follow-ups for PR #123: docs: list the cache env vars in server.json and .env.example Tracking auto-review follow-ups for #123 (verdict: **fail**). ### 🔴 Important - [ ] SKYLIGHT_TOKEN_CACHE description (112 chars) exceeds documented 100-character MCP Registry limit in server.json line 53 --- Checklist regenerated from the auto-review of `a24697a` ([run](https://github.com/chrischall/skylight-mcp/actions/runs/33125878355)). Ticks survive later rounds, but a round that reviewed an older commit can still re-list something already fixed — chec [Read the thread](https://github.com/chrischall/skylight-mcp/issues/124) · 2026-08-27 · closed · outside contributor · 1 comment ### Auto-review follow-ups for PR #121: feat(auth): accept a supplied refresh token via SKYLIGHT_REFRESH_TOKEN Tracking auto-review follow-ups for #121 (verdict: **fail**). ### 🔴 Important - [x] src/token-store.ts: the new `{ refreshToken }` CacheBinding arm (and its selecting ternary in src/auth.ts) has no covering test — tests/token-store.test.ts's binding tests only use {email,password}, and the new tests/auth.test.ts refresh-token tests bypass createTokenPersistence via persistence:null, so the branch is untested against the repo's documented 100% coverage gate [Read the thread](https://github.com/chrischall/skylight-mcp/issues/122) · 2026-08-26 · closed · outside contributor · 0 comments ### Auto-review follow-ups for PR #109: chore: declare a mint.yaml hosting manifest Tracking auto-review follow-ups for #109 (verdict: **warn**). ### 🟡 Nits - [ ] Redundant .mcpbignore entry for mint.yaml (already excluded by *.yaml pattern) --- Checklist regenerated from the auto-review of `57540d3` ([run](https://github.com/chrischall/skylight-mcp/actions/runs/32906126366)). Ticks survive later rounds, but a round that reviewed an older commit can still re-list something already fixed — check that SHA before redoing an item. When addre [Read the thread](https://github.com/chrischall/skylight-mcp/issues/110) · 2026-08-25 · closed · outside contributor · 0 comments ### Auto-review follow-ups for PR #95: fix(auth): find the CSRF token in any markup shape, and say why when it can't Tracking auto-review follow-ups for #95 (verdict: **fail**). ### 🔴 Important - [x] Untested branch: meta-tag content-before-name fallback regex in extractAuthenticityToken (src/auth-session-login.ts:101) is never exercised by any test, contradicting vitest.config.ts's branches:100 threshold and CLAUDE.md's 'Coverage gaps fail CI' rule ### 🟡 Nits - [x] status < 200 branch in describeTokenlessResponse (src/auth-session-login.ts:121) is also untested (all mock [Read the thread](https://github.com/chrischall/skylight-mcp/issues/96) · 2026-08-09 · closed · outside contributor · 0 comments ### Login fails at step 1: authenticity_token not found in /auth/session/new response skylight-mcp@0.4.6 login is failing for every call, with: Skylight login failed: could not find authenticity_token in /auth/session/new response. The Skylight login page may have changed. This is thrown from login() in [dist/auth-session-login.js](https://github.com/sarahgross1/personal/blob/main/dist/auth-session-login.js) (step 1 of the OAuth2 flow), before credentials are even submitted — so it reproduces regardless of SKYLIGHT_EMAIL/SKYLIGHT_PASSWORD correctness. Both env vars are confirmed [Read the thread](https://github.com/chrischall/skylight-mcp/issues/88) · 2026-07-30 · closed · external user · 1 comment ### Meal planning reads: no way to list planned meal sittings ## Gap `meals.ts` covers the meal-planning **writes** (issue #14 — `skylight_plan_meal`, recipe CRUD, `skylight_add_recipe_to_grocery_list`) but there is no **read** for what is actually on the meal plan. `skylight_list_recipes` returns the recipe box, not the schedule, so a question like "what is for dinner tonight?" cannot be answered by the server — the planned sittings are only visible on the frame itself. ## Endpoint `GET /frames/{f}/meals/sittings` serves exactly this. Verified live aga [Read the thread](https://github.com/chrischall/skylight-mcp/issues/86) · 2026-07-30 · closed · outside contributor · 0 comments ### Auto-review follow-ups for PR #69: fix: bump @chrischall/mcp-utils to 0.12.0 Tracking auto-review follow-ups for #69 (verdict: **warn**). ### 🟡 Nits - [ ] Use `dependencies` label instead of `bug` for dependency-only bumps (CLAUDE.md lines 139-152) --- When addressing the review on #69, resolve each item and add `Closes #` to the PR once everything is genuinely fixed; deferred items stay open here. See the auto-review follow-up convention in CLAUDE.md. [Read the thread](https://github.com/chrischall/skylight-mcp/issues/70) · 2026-07-07 · closed · outside contributor · 0 comments ### get_meal_sittings and get_meal_categories return "Unknown" instead of actual date/name data When calling get_meal_sittings, the tool returns meal sitting IDs but the date field is always "Unknown date" instead of the actual scheduled date: Scheduled meals: - Unknown date ID: 37563742 - Unknown date ID: 37563691 Similarly, get_meal_categories returns "Unknown" instead of the category name: Meal categories: - Unknown (ID: 1646939) - Unknown (ID: 1646940) For comparison, get_recipes works correctly and returns full recipe names, descriptions, and IDs without issue — so the bug appe [Read the thread](https://github.com/chrischall/skylight-mcp/issues/59) · 2026-06-30 · closed · external user · 1 comment ## Most recent The remaining reports are on [the project's issue tracker](https://github.com/chrischall/skylight-mcp/issues).