Playbook · operator how-to
How to source 10 dev-tool deals in a week (from GitHub trendlines)
A Tuesday-to-Friday playbook for sourcing ten qualified dev-tool deals using GitHub commit-velocity, contributor-influx, and star-detachment signals — no scraping required.
Sourcing ten dev-tool deals in five working days only feels hard because most pipelines start with cold scraping. The shortcut is to start from a pre-filtered signal feed, then spend your hours on qualification and outreach instead of data plumbing.
The playbook below is the exact cadence we run weekly on the GitDealFlow operating side. Tuesday is the data pull. Wednesday is qualification. Thursday is outreach drafting. Friday is the send. Total deep-work time across the week is roughly two and a half to three hours, with one hour of that being the Friday send window.
You can substitute any sector slug — ai-ml, fintech, security, data-infra — for devtools and the same cadence applies. The four-test qualification rubric (velocity-percentile, contributor-influx, star-detachment, infra-buildout) doesn't change.
Before you start
Prerequisites
- · An empty calendar block of 30–60 min on Tue, Wed, Thu, Fri
- · Read access to the public signals feed at /api/v1/signals.json (no auth required)
- · A simple outbound spreadsheet or CRM
Tools
- · GitDealFlow signals feed (free, no auth)
- · GitHub web UI (commit/contributor inspection)
- · Your CRM or a flat Google Sheet
Steps
- 01
Tuesday — pull the dev-tool sector slice
15mFetch
https://signals.gitdealflow.com/api/v1/signals.json?sector=devtools(no auth). You'll get the current week's top dev-tool repos ranked by composite signal score. Save the top 30 to a spreadsheet — slug, repo URL, current score, current commit-velocity percentile. - 02
Wednesday — apply the four-test rubric
45mFor each of the 30, check four signals: (1) commit-velocity 28-day percentile ≥ 80, (2) contributor-influx week-over-week ≥ +25%, (3) star-detachment (commits accelerating without star spike — the pre-attention zone) score ≥ 60, (4) infra-buildout (CI / containerization / dependency-graph density) score ≥ 50. Keep only repos passing 3 of 4. You should land in the 8–14 range.
- 03
Wednesday afternoon — strip out anything stage-mismatched
20mOpen each qualified repo's GitHub page. Cut anything where the README clearly says "hobby project", anything that's an obvious enterprise fork (microsoft/, google/, aws/), and anything where the top contributor has a corp email already (you can see this in the contributor list). Aim to land at ten.
- 04
Thursday — pull the warm-intro vectors
30mFor each of the ten, open the repo's contributor graph and the maintainer's profile. You're looking for (a) a Twitter / X handle in their bio, (b) a personal website, (c) a Discord/Slack community linked from the repo. The Twitter handle is your warm-intro vector — comment thoughtfully on one of their tweets first, send the DM second.
- 05
Thursday afternoon — draft the ten emails
40mUse a single template: line 1 references the specific commit-velocity signal ("saw your commit-velocity 28-day percentile jumped from 64 → 91 last week"), line 2 references one thing from the maintainer's bio or recent post, line 3 is one question about their thesis on the problem space. Keep each draft under 80 words.
- 06
Friday — send + log
30mSend all ten in one block. Log each into your CRM with the specific signal score that triggered the touch. Set a 14-day follow-up reminder. Don't expect more than 2–3 replies in the first 48 hours; the back-end conversion comes from the follow-up at day 14.
Run the play
See a worked example on /firstlook →Frequently asked questions
Do I need a GitDealFlow account to follow this playbook?
No. The /api/v1/signals.json feed is free and unauthenticated. The paid tier (Sector Sweep / Insider) speeds up the qualification step with pre-computed scoring exports, but the data is the same.
Can I substitute a different sector?
Yes. Replace `?sector=devtools` with any sector slug — `ai-ml`, `fintech`, `security`, `data-infra`, `dev-platforms`, `analytics`. The four-test rubric is sector-agnostic.
Why Tuesday and not Monday?
The signals feed refreshes Monday morning UTC. Tuesday gives the CDN cache time to warm globally and gives you 24 hours of post-refresh stability — fewer score deltas to chase.
What if I only have one hour per week, not three?
Skip Wednesday afternoon (the stage-mismatch sweep) and Thursday morning (the warm-intro vectors). You'll send eight cold-but-specific emails instead of ten warm ones. Reply rate drops ~40%, but you keep the cadence alive.