--- name: daily-radar description: Build a personal Daily Radar ("Your daily radar") dashboard of what shipped, what's moving, what's stuck, and what needs the user's attention, tailored to their role, team, and projects. Use for "daily radar", "what happened yesterday", "team status", "standup prep", "what's blocked", "what needs my attention", or "catch me up". --- # Your daily radar A morning briefing for one person. Read-only: never post messages, change tickets, or comment on PRs. ## How this runs Open with one friendly line and the plan: > 1. Work out what matters to you > 2. Gather the last day's activity > 3. Build your radar Start each step with a short progress line ("Step 2 of 3: ..."). Keep the tone plain. Ask at most one question, only when the answer changes the result. Finish with one suggested next step. ## Before you start This skill runs on the user's Devplan workspace. 1. Check that the Devplan tools are available. If your client loads tools on demand, search for "devplan" first. Never call a Devplan tool just to find out whether it exists. 2. If Devplan isn't connected, or a call fails with a sign-in error, say in one line that this needs Devplan, then ask setup's first question, "Do you have a Devplan account?", with its three options (**Yes, connect it**, **No, create one**, **What is Devplan?**), and follow `devplan-setup` from step 1. Stop this skill there. Don't give connect steps or commands of your own, and don't offer to do the task without Devplan (no local drafts, notes-based plans, or git summaries). 3. Load company and personal context with `devplan-product-context`. 4. If the workspace is still being built (Devplan is connected but `list_repository_descriptions` and `list_workspace_projects` both come back empty), work from what exists and don't pad. On a dashboard, show the template's notice ("Devplan is still analyzing your sources. This fills in automatically as it finishes."). Anywhere else, say it in one line. ## 1. Who it's for The radar is personal. Work out, in this order: 1. **Who they are.** From memory and earlier conversations: name, email, role, team. A scheduled run's instructions may name them and their focus; use that. 2. **What they care about.** The radar focus areas saved during setup or named in a scheduled run's instructions (product areas, teams, or themes, not project lists). Add from memory: customers they handle, people they work with, and anything they've said they want to watch. Map focus areas to work through the feature catalog and the projects and repos linked to those areas. 3. **Their footprint in Devplan.** Match their email to `list_workspace_members` and to a person in the graph. From the graph, find projects they own, PRs they authored or review, and work linked to their projects. Use this to order and filter, not to hide: their items come first, the team's important items follow. If you know nothing about them, show the team view and end with: "Want this tailored to you? Tell me your role and the areas to focus on." Don't save anything new to memory. ## 2. Window Default: the last 24 hours. On Monday, since Friday 00:00 local. Honor any window the user gives. ## 3. Gather Use `query_workspace_graph`, `search_entities`, and `list_workspace_projects`. Devplan has already attributed activity to people and projects across every source, so don't re-derive attribution. - **Shipped:** merged PRs and changes in the window, with their projects and features. - **Moving:** PRs opened, tickets started, projects with new linked activity. - **Stuck:** open PRs older than 2 days, tickets in progress with no activity for 3+ days, active risk insights, projects whose due date is close and completion is low. - **Talk:** discussion topics and meeting notes in the window with decisions, blockers, escalations, or customer issues. Skip chatter. - **For them:** PRs waiting on their review, new risks and slips in their focus areas and on work they own, mentions of their areas or customers in discussions and calls. - **Insights for their role:** active risk insights and open feature-request insights from the graph (`MATCH (i:Insight) WHERE (i.kind='risk' AND i.status='active') OR (i.kind='feature_request' AND i.status IN ['not_started','planned','in_progress'])`), filtered by `roleIds` to the user's role: founders and executives get `product_leader`, `engineering_leader`, `sales`, and `marketing`; product managers `product_manager`; engineering managers `engineering_manager`; and so on. Fetch them with `fetch_internal_entities_by_ids` (compact, a few at a time; large insights truncate) for their name, description, evidence, and suggested action. Rank by severity or business impact, then evidence count, then how recently they were updated. Example graph queries. Timestamps are epoch milliseconds: compute the window start from today's date and write the number in place of `$since`. ``` MATCH (p:Person)-[:AUTHORED]->(pr:PullRequest) WHERE pr.mergedAt >= $since RETURN p.name, pr.prId, pr.repo, pr.mergedAt ORDER BY pr.mergedAt DESC LIMIT 100 MATCH (x)-[:LINKED_TO]->(proj:Project) WHERE (x:Change OR x:Signal) AND x.periodEnd >= $since RETURN proj.projectId, count(x) AS activity, max(x.periodEnd) AS last ORDER BY activity DESC LIMIT 50 ``` If a date-filtered `search_entities` call comes back empty, search again without the date filter and keep results whose source date falls in the window; recent sources often aren't indexed by source time. People who wrote PRs but aren't in `list_workspace_members` have left. Their open PRs are a Needs your attention item (decide: close, hand off, or finish), not noise. Project nodes have no titles, and their graph status can lag; take titles, status, and owners from `list_workspace_projects`, joined by ID. People in the graph are often named by their git username, so match them by email or `gitUsername`, and show the name from `list_workspace_members` when you can. Leave out bots (names ending in `[bot]`, dependabot, renovate). ## 4. Dashboard Title: "Your daily radar". Masthead scope: "For " when personalized, otherwise "Team view". No figures. 1. **Summary:** 3 to 5 concise bullets with the key updates since the last radar, led by what changed in the themes below. Plain text, no links. 2. **Needs your attention:** 3 to 6 themes scoped to the user's role and focus areas, about the bigger things: revenue and customers, product direction, launches, security and trust, team capacity. Build them from the role-filtered insights, then fold in what the last day added across customer calls, internal discussions, tickets, and code. One row per theme: a short name, one or two sentences on what's at stake and what changed recently, the owner if there is one, evidence (the insights' sources, calls, threads, PRs), and **always** one action button that moves the theme forward. Use the insight's own suggested action prompt when it has one; otherwise write one (Draft brief, Draft message, Plan coverage, Write spec). Roll small items (one stale PR, one missing review) into the theme they belong to or leave them out. A manager sees team-level themes; an individual contributor sees their own work. If nothing needs attention, use the Empty pattern. 3. **Product updates:** what changed in the product in the window, as dated entries grouped by area. Each entry names the specific change and says how it affects users, in plain language ("Assistants connected through MCP can now save a spec straight into a Devplan project, so specs written in Claude show up for the whole team"). Group internal-only work (dependency bumps, refactors) into one line that says there's no user-facing change. Mark changes not yet analyzed as such. 4. **In motion:** a group per active project, with its new activity as child rows. Every item has its evidence behind the Evidence link. ## Dashboard format Build the page from [references/dashboard-template.md](references/dashboard-template.md). Don't write your own layout, CSS, or script. 1. Copy the template's ``, `