db-aeon/joshu-oss/last30days-gui
Claude Code and agent skill by db-aeon/joshu-oss
- community
Checking that the SKILL.md file is still in the repository.
last30days GUI — research/Watching with the app window open.
Install
npx vskill@latest install db-aeon/joshu-oss/last30days-guiSKILL.md
last30days GUI skill
Use when the last30days window is open (embedded chat or voice). GUI-first: drive the open UI. Escalate to plugin tools only when the owner is in jChat / Telegram with the app closed.
Platform boundary
| User intent | App window open | App closed (jChat / Telegram) | |
|---|---|---|---|
| Run research | `runResearch` { topic, days?, depth? } — blocks until done | `last30days_research` — async; Joshu replies in chat | |
| Cancel in-flight run | `cancelRun` | — | |
| Open prior run | `openRun` { runId } or snapshot `recentRuns` | read joshu://research/last30days/… | |
| Settings / doctor | `openSettings` / `runDoctor` | ask owner to open Settings (keys live there) | |
| Switch to Watching | `openWatching` | `last30days_watch_list` | |
| Watch a topic | `watchThisTopic` { topic? } | `last30days_watch_add` then `last30days_watch_run` | |
| Check watches now | `openWatching` (owner clicks Check now) | `last30days_watch_run` / `last30days_watch_run_all` | |
| Watch report | Watching row in the UI | `last30days_watch_report` { topic } | |
| Stop watching | Watching screen | `last30days_watch_remove` { topic } |
Do not reply “starting research” before `runResearch` returns. Do not use app_gui_action from jChat when the window is closed. Do not invent raw HTTP; plugin tools wrap invoke.
Joshu policy: ScrapeCreators for social/YouTube; X via xquik (fleet relay or self-host key). No browser cookies, yt-dlp, or XAI. Web via Exa when provisioned else keyless DuckDuckGo.
Query planning is automatic — pass only the topic to runResearch / invoke research. Joshu builds the engine QueryPlan server-side (named entity vs concept vs comparison). Do not mention QueryPlan to the owner.
A report is a one-shot Research run (no trending-vs-average). A watch is an explicit Watching add; trending needs ≥3 watch snapshots. Cron jobs last30days: watchlist daily (08:00) and … weekly (Mon 08:00) are Hermes Schedules no-agent scripts, not Linux crontab.
Output files
Every completed run writes markdown under `research/last30days/` (gbrain-indexed):
- Path is relative to
${JOSHU_FILES_ROOT}— e.g.joshu://research/last30days/2026-08-07-cod-a1b2c3d4.md - YAML frontmatter:
type: research,source: last30days,topic,run_id, stats - Use for email drafts, voice summaries, or
read_filein Hermes
GUI snapshot fields
- `activeView`: research | watching
- `topic`: current topic field
- `activeRunId`, `runStatus`: live job if any
- `resultPreview`: short summary of Results panel
- `recentRuns`: last few run ids + statuses + topic + when
- `engineReady`: engine present on box
When `runResearch` returns, summarize key findings and mention the saved joshu://… report path.
Headless invoke (plugin tools wrap these)
| action | Purpose | |
|---|---|---|
research | { topic, days?, mock?, wait?: boolean, hermesSessionKey? } — async by default | |
watchingList | List topics + trending | |
watchingAdd | { topic, cadence? } — no research | |
watchingRemove | { topic } | |
watchingReport | { topic } | |
watchingRun | { topic } — async 7d recheck | |
watchingRunAll | { cadence? } — async | |
watchlistRunAll | Blocking cron path | |
doctor | Health JSON | |
briefingGenerate | Generate briefing from store |
Deep web outside loaded results: skill_view('joshu-browser') when needed.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingarozos/subservice/last30days/skills/last30days-gui/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/db-aeon/joshu-oss/last30days-gui)