Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/brownglasses/dotplot-mcp/dotplot-whats-changednpx skills add brownglasses/dotplot-mcp --skill dotplot-whats-changedgit clone --depth 1 https://github.com/brownglasses/dotplot-mcpWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00086 | $0.00566 |
| Opus 5 | $0.00043 | $0.00283 |
| Sonnet 5 | $0.00017 | $0.00113 |
| Haiku 4.5 | $0.00009 | $0.00057 |
Grade A, and why
dotplot-whats-changed scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
What changed since last time
Snapshots are saved locally to ./.dotplot/history.json every time a report is
made. This reads them.
1. Use the same data source and the same value event
A comparison is only meaningful if both sides were measured the same way. Pull
the events the same way as before (same query, same tables), and pass the same
value_event — switching from purchase to signup between runs produces a
dramatic-looking change that means nothing.
Call history_compare, or analyze (which saves a new snapshot and shows the
comparison card in the report).
2. Trust the tool when it says there's nothing to compare
Only snapshots of the same dataset are compared — matched by a fingerprint of the user IDs, not by the file name or the event name. So:
- First run on this product →
notesays there's no history yet. Say that plainly. Don't imply nothing changed. - Analyzed a different product from the same folder → it won't be compared, on
purpose. Two products both using the event name
purchaseare not the same product.
3. Read the size of the move, not just the direction
Every rate in the comparison has a denominator. With 30 users, one person changes churn by 3 points — a "5 point improvement" is two people changing their minds. State the user counts alongside the percentages.
A move is worth acting on when it's larger than a few users' worth of noise, and when you can point at something that changed in the product to explain it.
4. Connect it to what they did
The useful version of this answer is not "regular_rate went from 8% to 14%". It is "the onboarding change you shipped on the 5th is followed by more users reaching a second week — worth continuing, though it's 25 users so far."
If they shipped nothing in between, say that too. Metrics drift on their own, and reading meaning into that drift is how teams end up chasing noise.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 60 lines · 86 tokens per session scan A 8d1ce8c48983
dotplot-whats-changed is a skill published in the GitHub repository brownglasses/dotplot-mcp (1 stars, last pushed 19d ago), licensed MIT. It adds 86 tokens to every session and 566 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
data-analysis
业务数据分析助手（电商/增长方向）。连接 MySQL 数据库或读取 CSV/Excel，完成需求澄清 → SQL 查询 → 数据体检 → 全自动清洗 → 自动 EDA 探索 → 漏斗/留存/复购/RFM/指标异动归因分析 → 可视化 → 结论先行分析报告的完整流程；内置只读安全防护、MySQL 8 SQL 模板、10 个开箱即用脚本和演示电商数据，对新手友好。当用户提到分析数据、查数据库、写 SQL、漏斗、留存、转化率、GMV、DAU、复购、用户分群、指标为什么涨/跌、做报表，或给出任何 csv/xlsx/数据库数据想找业务结论时都应使用——用户不必说出"数据分析"四个字。.
call-coach
B2B discovery call coach. Paste any transcript — Granola export, Fireflies, Fathom, plain text, a file path — and get a full coaching report: framework flags, gaps the rep failed to build, admissions captured, and a complete next-call script. Uses local memory files to track deals across multiple calls so coaching…
cohort-analysis
Build a full SaaS cohort analysis from a CRM (Attio/Stripe/CSV) joined to revenue (Stripe/Attio/CSV). Outputs a styled Excel workbook with conditional formatting (Customer Churn, MRR Churn, CAC Payback) plus per-section CSVs. Use when the user says "/cohort-analysis", "build a cohort table", "cohort analysis for "…
attio-deepline-enrich
Enrich records inside Attio with a clear cost gate before any credit is spent. Pull a chosen object (companies, people, or custom) from Attio, fill chosen attributes via Deepline providers (emails, phones, firmographics, LinkedIn, ICP fields), show the exact credit and dollar cost, wait for an explicit yes, run it…
cfo-skill
Read-only CFO data dashboard for bootstrapped startups. Two modes, CSV templates (works with any stack: Attio/HubSpot/Salesforce + Qonto/Mercury/Brex + Stripe + Moss/Ramp/Pleo) or live API pull from Attio + Qonto + Stripe + Moss. Computes runway, burn rate, MRR/ARR, NRR, customer concentration, AR aging, DSO, vendor…
linkedin-self-improvement-loop
A build-measure-learn loop for your LinkedIn. Ingests your Creator analytics export, keeps a persistent belief model of what drives your reach and engagement, reconciles last cycle's beliefs against the new data, proposes ONE experiment to run next, and hands draft briefs to a drafting skill. Run it on a cadence and…