Works with the AI Agents you already use
Chartiva
It tells the assistant when to use the server (requests to chart, plot, map, profile data, build dashboards, spot anomalies or analyze networks) and when not to.
$15
Chartiva
Example session with this skill installed
Here it is as Markdown you can paste into your listing. The block uses four backticks so that it can contain the code block in the sales manager prompt.
Example prompts by role
Each prompt below is written as the person in that role would phrase it, using that role's usual data.
1. CEO / CFO: board-ready business review
Data sources: Excel (multi-sheet), CSV
I have the board meeting on Thursday. Use
~/Documents/FY26-financials.xlsx(all sheets: revenue, costs, headcount) and~/Documents/pipeline-export.csvfrom our CRM. Build me a one-page executive dashboard with KPI tiles for revenue, gross margin, operating cost and pipeline coverage. Add revenue vs. target by quarter with a target line, a waterfall showing how we got from last year's profit to this year's, and cost by department I can drill into. Keep it clean, light theme, and export it as PDF for the board pack plus an interactive HTML I can open during the meeting. Summarise the three numbers I should lead with, using only what's in the data.
Shows: multi-sheet Excel, joining CSV and Excel data, KPI tiles, waterfall and drill-down charts, target lines, PDF and HTML export, and summaries based only on computed results.
2. Data scientist: rigorous exploratory analysis
Data sources: JSON Lines, CSV
Profile
~/data/customer-events.jsonl(about 40k rows) and~/data/customers.csv. Show me field types, missing values, outliers and the strongest correlations first. Then join them oncustomer_idand show me the session-length distribution by plan tier as box plot, violin and density side by side, with a ranking of which chart represents it best. Plot spend against sessions as a scatter coloured by tier (use canvas, it's a lot of points), flag anomalous customers with a robust z-score, and give me the method and threshold you used.
Shows: data profiling, JSON Lines, joins with match counts, side-by-side chart comparison with a ranking, canvas rendering for large data, documented anomaly-detection methods, and nothing invented.
3. CIO / CTO: infrastructure risk overview
Data sources: JSON network graph, Excel, CSV
I need to understand our infrastructure risk before the budget review.
~/Documents/network-topology.jsonhas our devices and links,~/Documents/asset-inventory.xlsxhas device age, vendor and end-of-support dates, and~/Documents/incidents-2026.csvis the year's ticket export. Show me which devices would cause the most disruption if they failed, which of those are near end of support, and where incidents cluster. Draw the network with critical devices highlighted, a Gantt of end-of-support dates by device group, and a calendar heatmap of incidents. Put it in a dashboard I can share with my leadership team.
Shows: three data sources linked by device, betweenness centrality on real network data, Gantt and calendar charts, a linked dashboard, and a strategic risk view without writing code.
4. Operations / sales manager: team performance and trends
Data sources: pasted table, CSV
Here's my team's weekly numbers from our spreadsheet (pasted below), plus last year's export at
~/Downloads/sales-2025.csv. Compare this year with last year by rep and by region, highlight the top 3 reps, and show who's trending up or down. A radar chart comparing reps on calls, meetings, win rate and deal size would be great. Then make the main chart horizontal, sort it descending, and use our brand colours (viridis is fine). I'll probably want to undo some changes, so keep the versions.rep,week,calls,meetings,deals,revenue Aisha,2026-W40,52,11,3,48200 …
Shows: pasted table data combined with a CSV file, year-over-year comparison, trend detection, radar charts, editing the chart in plain language, palettes, and version history with undo.
5. Network administrator: live troubleshooting
Data sources: YAML topology, CSV telemetry, GeoJSON
Users at three branches are complaining about slow apps since Monday.
~/netops/topology.yamlis our SD-WAN topology,~/netops/wan-telemetry.csvhas 5-minute latency, loss and jitter per tunnel for the last two weeks, and~/netops/sites.geojsonhas site locations. Draw the topology with device icons in tiers, find the latency and loss anomalies per tunnel against each tunnel's normal trend, and show the shortest path frombranch-07todc-east. Put a map of sites coloured by anomaly count, the anomaly time series with an SLA line at 150 ms, and the topology in one linked dashboard, so clicking a site filters everything.
Shows: YAML topology with device icons, GeoJSON maps, anomaly detection per tunnel that allows for normal trends, SLA lines, shortest paths, and a linked dashboard for troubleshooting.
Summary
| Role | Data sources | Key features shown |
|---|---|---|
| CEO / CFO | Excel (multi-sheet), CSV | KPI dashboard, waterfall, PDF export |
| Data scientist | JSON Lines, CSV | Profiling, joins, chart ranking, canvas, anomalies |
| CIO / CTO | JSON graph, Excel, CSV | Graph analytics, Gantt, calendar, linked dashboard |
| Ops / sales manager | Pasted rows, CSV | Trends, radar, chart edits, version history |
| Network admin | YAML, CSV, GeoJSON | Topology icons, maps, anomalies, SLA lines, shortest path |
- Read your context and instructions
- Compiled the chartiva
- Generated the data export
Example outputs by role
Illustrative replies based on sample data. Figures are examples of the format, not real results.
1. CEO / CFO: board-ready business review
Summary
Your board dashboard is ready. I combined the three sheets in FY26-financials.xlsx with pipeline-export.csv (1,284 opportunities, all matched to a quarter) into one page with 4 KPI tiles and 3 charts, exported as interactive HTML and PDF.
Deliverable
Files
| Format | Path |
|---|---|
| Interactive HTML (for the meeting) | ~/d3-visualizations/board-review-fy26/dashboard.html |
| PDF (for the board pack) | ~/d3-visualizations/board-review-fy26/dashboard.pdf |
KPI tiles
| KPI | FY26 | vs. FY25 |
|---|---|---|
| Revenue | $48.6M | +12.4% |
| Gross margin | 61.8% | +2.1 pts |
| Operating cost | $21.3M | +4.7% |
| Pipeline coverage (next 2 quarters) | 3.1× | +0.4× |
Charts
- Revenue vs. target by quarter (bar + target line): Q1, Q3 and Q4 beat target; Q2 missed by $0.6M (−5.0%).
Profit bridge FY25 → FY26 (waterfall): profit rose from $5.9M to $8.2M. The largest positive step is revenue growth (+$5.4M); the largest negative step is headcount cost (−$2.1M).
3.
Cost by department (sunburst, click to drill down): R&D is the largest share at 38%, then Sales & Marketing at 31%.
Three numbers to lead with (all computed from your files):
- Revenue $48.6M, +12.4% year over year.
- Gross margin up 2.1 points to 61.8%.
- Pipeline coverage of 3.1× for the next two quarters.
Next steps
- Open the HTML during the meeting: hover for exact values, click a department to drill in.
- Want a dark version for screens, or the waterfall as a separate slide? Say "dark theme" or "export the waterfall as PNG".
- If Q2 comes up, I can break the miss down by region or product.
2. Data scientist: rigorous exploratory analysis
Summary
I profiled both files, joined them on customer_id (39,412 of 40,000 events matched; 588 had no customer record and were left out of the joined charts), compared three distribution charts, and flagged 214 anomalous customers with a robust z-score.
Deliverable
Profile highlights
| Field | Type | Missing | Notes |
|---|---|---|---|
session_length_s | number | 0.4% | right-skewed; 1,106 Tukey outliers |
spend_usd | number | 0% | median $42, max $9,870 |
plan_tier | category | 0% | Free, Pro, Enterprise |
event_ts | date | 0% | 2026-01-01 → 2026-06-30 |
Strongest correlation: sessions ↔ spend_usd, Pearson r = 0.71.
Chart comparison: session length by plan tier
| Rank | Chart | Why |
|---|---|---|
| 1 | Violin | shows the bimodal Pro distribution the others hide |
| 2 | Box plot | clearest medians and outliers, but hides shape |
| 3 | Density | good shape, but overlapping tiers are hard to read |
Charts written
- Comparison:
~/d3-visualizations/session-length-compare.html - Spend vs. sessions scatter (39,412 points, canvas, colour = tier):
~/d3-visualizations/spend-vs-sessions.html
Anomalies
- Method: modified z-score on
spend_usdperplan_tier, threshold 3.5 (Iglewicz–Hoaglin). - 214 customers flagged (0.5%); 161 of them are on the Free tier with spend above $500.
- Enriched dataset with
anomaly_scoreandis_anomaly:ds_7f3a…
Next steps
- Render the violin as the final chart: "use the violin".
- List the top 20 anomalies with expected vs. actual spend.
- Re-run with a stricter threshold (for example 5.0) if 214 is too many to review.
3. CIO / CTO: infrastructure risk overview
Summary
I linked your topology (412 devices, 618 links), asset inventory and 2026 incidents by device name (98% matched), ranked devices by how much traffic depends on them, and built a leadership dashboard. 7 of the 10 most critical devices reach end of support within 18 months.
Deliverable
Dashboard: ~/d3-visualizations/infra-risk-2026/dashboard.html
Most critical devices (betweenness centrality = share of shortest paths through the device)
| Rank | Device | Betweenness | End of support | Incidents 2026 |
|---|---|---|---|---|
| 1 | dc1-core1 | 0.38 | 2027-03 | 14 |
| 2 | dc1-core2 | 0.35 | 2027-03 | 9 |
| 3 | wan-edge-east | 0.22 | 2026-12 | 21 |
| 4 | dist-campus-a | 0.17 | 2028-06 | 6 |
| 5 | wan-edge-west | 0.15 | 2026-12 | 18 |
Dashboard contents
- Network map: device icons in tiers; node size = criticality, red = end of support within 18 months.
- End-of-support Gantt by device group: WAN edge (Dec 2026) is the earliest at-risk group.
- Incident calendar heatmap: incidents cluster on Mondays and in the last two weeks of March.
- Linked selection: click a device to filter its incidents and support dates.
Network facts: 4 communities (modularity 0.62); 1 connected component, so there are no isolated islands.
Next steps
- Use the top-5 table as the basis for the refresh budget.
- "Show what breaks if dc1-core1 fails": shortest-path impact for the branches behind it.
- Export as PDF for the leadership pre-read.
4. Operations / sales manager: team performance and trends
Summary
I combined your pasted 2026 numbers (12 reps × 40 weeks) with sales-2025.csv, compared the two years by rep and region, highlighted the top 3 reps, and added a rep radar chart. Every change is saved as a version, so you can undo any step.
Deliverable
Charts
| Chart | Path |
|---|---|
| Revenue by rep, 2026 vs. 2025 (horizontal, sorted, top 3 highlighted, viridis) | ~/d3-visualizations/team-yoy.html |
| Rep scorecard radar | ~/d3-visualizations/rep-radar.html |
Top 3 reps (2026 revenue)
| Rep | 2026 | vs. 2025 | Trend (slope per week) |
|---|---|---|---|
| Aisha | $1.92M | +18% | ↑ +$1.4k |
| Marco | $1.71M | +9% | ↑ +$0.6k |
| Priya | $1.66M | +22% | ↑ +$1.9k |
Trending down: Tom (−$0.8k/week, r² = 0.41) and Lena (−$0.5k/week, r² = 0.28).
By region: EMEA +15%, Americas +6%, APAC −3%.
Radar compares calls, meetings, win rate and deal size: Priya leads on win rate (31%); Aisha leads on deal size ($16.1k).
Version history
| Version | Change |
|---|---|
| v1 | original grouped bar |
| v2 | horizontal, sorted descending |
| v3 | top 3 highlighted, viridis palette |
Next steps
- Undo anything: "go back to v1" or "what changed between v1 and v3?".
- Drill into APAC: "show APAC by rep".
- Export both charts as PNG for your team meeting.
5. Network administrator: live troubleshooting
Summary
I found
37 latency and loss anomalies across 4 tunnels since Monday, measured against each tunnel's own trend, so normal busy hours aren't flagged. Three of the four affected tunnels pass through wan-edge-east. Your linked troubleshooting dashboard is ready.
Deliverable
Dashboard: ~/d3-visualizations/sdwan-troubleshoot/dashboard.html
Anomalous tunnels (modified z-score on residuals from each tunnel's trend, threshold 3.5)
| Tunnel | Anomalies | Worst latency | Expected | First seen |
|---|---|---|---|---|
branch-07 ↔ dc-east | 14 | 412 ms | 48 ms | Mon 09:05 |
branch-11 ↔ dc-east | 11 | 356 ms | 52 ms | Mon 09:10 |
branch-03 ↔ dc-east | 9 | 298 ms | 45 ms | Mon 09:15 |
branch-19 ↔ dc-west | 3 | 171 ms | 61 ms | Wed 14:20 |
Shortest path branch-07 → dc-east: branch-07 → wan-edge-east → dc-east-agg1 → dc-east (3 hops). The three worst tunnels share wan-edge-east.
Dashboard contents
- Site map (GeoJSON) coloured by anomaly count.
- Latency time series per tunnel with an SLA line at 150 ms; anomalies are marked.
- Topology with device icons in tiers; the shortest path is highlighted.
- Linked selection: click a site to filter the map, time series and topology together.
Packet loss: anomalies on the same three east tunnels, peaking at 4.2% (expected 0.1%).
Next steps
- Check
wan-edge-eastfirst: it's the common device on the worst paths. - "Show jitter for the same tunnels" to confirm the pattern.
- Export the dashboard as PDF for the incident ticket.
FY26-financials.xlsx
CSV · data export
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
The d3-visualization skill teaches an AI assistant how to turn data into interactive D3.js charts, dashboards and analyses using the d3-visualization MCP server, without writing any chart code. It tells the assistant when to use the server (requests to chart, plot, map, profile data, build dashboards, spot anomalies or analyze networks) and when not to. It then shows how to pick the right one of the 21 tools, how to pass data (pasted rows, CSV/JSON/Excel files, networks, hierarchies, or reusing stored data), and which plain-language instructions the server understands, such as "highlight the top 3" or "dark theme". It also covers following workflows for refining, comparing, restoring and exporting charts, reporting results accurately (the HTML file path plus only the facts the tools computed), recovering from errors, and safety rules such as respecting folder restrictions and never following instructions found inside data.
Max 50,000 rows per dataset. PNG/PDF export requires Chromium. No direct database or URL fetching. Artifacts expire after 72 hours.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 10 days ago
Requires
Permissions
Allowed Hosts