# Daily Memory 2026-04-14

## Day recap 2026-04-13

- **What happened:**
  - **Email digestion & prioritization:** The agent continued summarizing a high volume of automated emails (Naver Pay purchase/confirmation notices, Saramin job recommendations, LinkedIn suggestions, Green Man Gaming promotions, bank/open-banking notifications, Apple iCloud+ receipt). Each was tagged by urgency (critical security, finance, or career vs low-priority marketing/noise).
  - **Security-relevant events:** Flagged a Google security alert for a new Windows login on `choonarm5@gmail.com`, multiple financial emails (NICEPAY/KB, Naver Pay, electronic tax invoice from Anyline, KB open-banking data-sharing notice), and several login/verification codes (PTKOREA, Jeju Bank).
  - **Career/job search signals:** Surfaced and tracked deadlines for key Saramin postings (Supercent AI Application Engineer, Kia AI/image analysis internship, AI/backend roles at EigenDrug, NextCore, LeadNumber, CL Mobility, and other curated job lists and public-sector opportunities).
  - **Dev/CI monitoring:** Monitored GitHub Actions failures for `LLagoon3/Portfolio-Project` (API ESLint config missing) and `LLagoon3/ai-usage-agent` (test job failure) and confirmed a successful `portfolio-project-v0.1.1` release and later CI/test hardening in `ai-usage-agent`.
  - **Tooling work (Claude/Codex/agent stack):** Subagents iterated on `ai-usage-dashboard` and `ai-usage-agent`, establishing a clean CLI agent package layout, Codex usage adapter, test suite and CI workflow, plus an initial Claude auth/account-resolution layer with doctor/status CLI integration.

- **Decisions / stable facts:**
  - `ai-usage-agent` is locked in as an independent npm-installable CLI/agent, with `provider-adapters`, `schemas`, and `shared` kept as internal dependencies.
  - For CI in `ai-usage-agent`, use Node's built-in `node:test` with `npm install` (not `npm ci`) until a root lockfile exists.
  - Claude auth work will continue via **importing existing Claude CLI credentials** and prioritizing `agent-store > claude-cli-import > not-found`, with changes applied in very small, test-verified slices.
  - Ongoing main-session priorities include: securing Google accounts, validating significant financial transactions/invoices, and addressing CI lint/test failures.

- **Next actions / blockers:**
  - **Security:** Verify the legitimacy of the new Windows login on `choonarm5@gmail.com`; if suspicious, change the password and review recent activity.
  - **Finance:** Double-check large or unusual financial items (Anyline 500k KRW invoice, 810k KRW HP monitor purchase, KB open-banking data-sharing notice) via official apps/sites instead of email links.
  - **Career:** Decide which near-deadline roles to actually apply to (Supercent AI Application Engineer by 4/15, Kia AI internship by 4/20, plus selected AI/backend roles from Saramin digests).
  - **Dev/CI:** Finish fixing `apps/api` ESLint configuration in `Portfolio-Project` and address the failing `test` job in `ai-usage-agent` if any remaining red builds appear.
  - **Auth/tooling:** Complete `gog` PC-browser authentication and continue incremental Claude auth integration in `ai-usage-*` repos without introducing live write flows yet.

- **Links/IDs:**
  - Daily memory auto-summary cron id: `6ef3619b-e82b-4c0c-b39c-7973eaf01422`.
  - Portfolio-Project release: `portfolio-project-v0.1.1`.
  - Naver Pay large purchase: order `2026041337210501` (HP monitors, 810,000 KRW).
  - Anyline tax invoice document: `20260413104120SA2LY8WCAIPHRBLGQQKSV0HI8AELOVSN`.
  - Gmail security alert account: `choonarm5@gmail.com`.

## Auto-summary 2026-04-14 10:00 KST

- **What happened:**
  - **Newsletter Analysis:** Summarized "TheV씨" newsletter regarding 2025 financial performance of AI and E-commerce startups; noted that most AI startups remain in deficit (except Moloco) while E-commerce shows selective turnarounds (Danggeun, Kurly).
  - **CI Monitoring:** Flagged a new GitHub Actions failure in `LLagoon3/ai-usage-agent` on the `feat/claude-auth-foundation` branch (job `test` failed).
  - **Notion Documentation:** Tracked updates by SW to "Mail server replacement" and "Mail error response" pages, including new `Test-NetConnection` troubleshooting commands and administrator links.
  - **Career Signals:** Digested morning Saramin recommendations, highlighting upcoming deadlines for Tokyo Electron (4/26), Hyundai Glovis internship (4/18), and Gyeongbo Pharm (4/19).
- **Decisions / stable facts:**
  - Confirmed that `ai-usage-agent` development on the Claude auth layer is currently blocked by failing tests in CI.
- **Next actions / blockers:**
  - **Dev:** Investigate and fix the failing `test` job on `ai-usage-agent` (`feat/claude-auth-foundation` branch).
  - **Career:** Review Hyundai Glovis (4/18) and Gyeongbo Pharm (4/19) postings before their weekend deadlines.
- **Links/IDs:**
  - ai-usage-agent failing run: `https://github.com/LLagoon3/ai-usage-agent/actions/runs/24370266274`.
  - Updated Notion page: "메일 송수신 오류 대응" / "메일서버 교체에 따른 변경 사항".

## Auto-summary 2026-04-14 13:00 KST

- **What happened:**
  - **Newsletter/Ad Digestion:** Processed Naver Cloud Platform's invitation to a "Public AX Strategy Seminar" in Sejong scheduled for April 23rd.
  - **Testing:** Observed a successful delivery of a self-sent test email from `lagoon3@everdigm.com`.
  - **Brand Monitoring:** Digested T'way Air's announcement regarding its planned name change to "Trinity Airways," noting that the rebranding is subject to regulatory approval and current operations remain unchanged.
- **Decisions / stable facts:**
  - T'way Air will continue to operate under its existing brand and booking systems until official approval for "Trinity Airways" is finalized (expected to take several months).
- **Next actions / blockers:**
  - No new blockers identified, continue monitoring career-related deadlines and CI status reported in the previous period.

## Auto-summary 2026-04-14 16:00 KST

- **What happened:**
  - **Job Monitoring:** Processed a Jobkorea notification for a short-form video creator position at "Hot Seller" (핫셀러), an IT consulting firm with 12.3B KRW revenue and 80 employees.
  - **Career Signals:** Digested a large Saramin recommendation batch (15 roles), prioritizing urgent deadlines for **EBM Engineering** (AI Developer) and **Incubus** (Full-stack Developer) which are closing today (D-0). Other noted roles include 2CSG, EigenDrug, and Nextcore Technology.
  - **System Stability:** Encountered a transient "ChatGPT usage limit" error for the gpt-5.4 model at 11:02 KST, but session recovery was successful and subsequent hooks were processed normally.
- **Decisions / stable facts:**
  - Identified EBM Engineering and Incubus as the highest priority for career review today due to immediate (D-0) deadlines.
- **Next actions / blockers:**
  - **Career:** Review and potentially apply to EBM Engineering (AI Dev) and Incubus (Full-stack Dev) before the end of the day.
- **Links/IDs:**
  - Urgent Saramin roles: EBM Engineering (AI), Incubus (Full-stack).

## Auto-summary 2026-04-14 20:00 KST

- **What happened:**
  - **Repetitive CI Failures:** Monitored a series of GitHub Actions failures in the LLagoon3/ai-usage-agent repository across multiple branches (feat/claude-usage-foundation, feat/claude-oauth-foundation, feat/claude-auth-foundation, and dev).
  - **CI Root Cause Identified:** Determined that the test jobs are consistently failing during the actions/setup-node@v4 step because the workflow is configured to use cache: npm but cannot find a package-lock.json (or similar lock file) in the repository root.
  - **LinkedIn Activity:** Processed a low-priority notification regarding a Malwarebytes milestone shared on LinkedIn.
- **Decisions / stable facts:**
  - The ai-usage-agent CI workflow requires a fix either by providing a root lock file or adjusting the setup-node caching configuration (e.g., specifying a cache-dependency-path or disabling caching until a lock file is committed).
- **Next actions / blockers:**
  - **Dev/CI Fix:** Resolve the setup-node caching issue in the ai-usage-agent GitHub Actions workflow to unblock testing across all feature branches.
- **Links/IDs:**
  - ai-usage-agent CI run (representative): https://github.com/LLagoon3/ai-usage-agent/actions/runs/24395110977

## Auto-summary 2026-04-14 20:00 KST (second run)

- **What happened:**
  - No new main-session activity (`agent:main:main`) occurred between the earlier 20:00 summary and this run; no additional files, configs, or jobs were changed.
- **Decisions / stable facts:**
  - All decisions and CI root-cause findings remain exactly as recorded in the prior 20:00 summary.
- **Next actions / blockers:**
  - Continue with the CI workflow fix for `LLagoon3/ai-usage-agent` as previously noted; no new blockers or tasks were introduced in this interval.
- **Links/IDs:**
  - Daily memory auto-summary cron id (20:00 announcer): `98d4edb5-9e53-4892-a788-58ab101c8b95`.

## Day recap 2026-04-14

- **What happened:**
  - Processed an Apple receipt email confirming renewal of the ChatGPT Plus monthly subscription (billed via Apple on 2026-04-13, ~29,000 KRW including VAT). No issues or disputed charges were indicated.
- **Decisions / stable facts:**
  - ChatGPT Plus remains active and is billed through Apple with the next renewal scheduled for mid-May 2026.
- **Next actions / blockers:**
  - Optionally review recurring subscriptions around mid-May if there is any desire to adjust or cancel ChatGPT Plus before the next billing cycle.
- **Links/IDs:**
  - Service: ChatGPT Plus (monthly) receipt from Apple for April 2026.