fork-fleet

Inventories active GitHub forks, scores diverged work, and surfaces upstream contribution candidates.

Updated May 25, 2026
One-click install
npx skills add https://github.com/Cbowcrptex/CBOWCRYPTEX --skill fork-fleet-cbowcrptex
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: fork-fleet
Source: https://github.com/Cbowcrptex/CBOWCRYPTEX/tree/main/skills/fork-fleet
Command: npx skills add https://github.com/Cbowcrptex/CBOWCRYPTEX --skill fork-fleet-cbowcrptex

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Teams maintaining open-source projects lose visibility into what their fork ecosystem is building, missing valuable community contributions that could be merged upstream and wasting time manually checking dozens of forks. ## Core Features & Use Cases - Fork Inventory & Classification: Lists all forks via the GitHub API and classifies them as active, stale, or dormant based on push activity. - Divergence Scoring: Compares each active fork against the parent repo in a single API call, then weights signals like new skills, modified files, and unique commits into PROMOTE, REVIEW, or NOTE tiers. - Gated Weekly Reports: Writes a structured markdown report with a one-line verdict and week-over-week delta, sending notifications only when something actually changed. - Use Case: A maintainer of an agent framework schedules this weekly to discover that a community fork added three new skills, reads the synthesized summaries, and opens a PR to cherry-pick the best one upstream. ## Quick Start Run the fork-fleet skill to scan all active forks of this repository and generate a report of upstream contribution candidates.

Frequently Asked Questions about fork-fleet

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I track what forks of my GitHub repo are building?▼

Use the GitHub API to list forks with their pushed_at timestamps, then run a cross-repo compare of each active fork against the parent default branch. This returns ahead/behind counts, unique commits, and changed files in a single call per fork.

How to find upstream contribution candidates from GitHub forks?▼

Score each active fork by weighting substantive changes like new files under skills/ higher than config tweaks, then deep-read the top candidates' new files. Forks with new skill files become PROMOTE candidates worth cherry-picking into a PR.

Does this work with private or archived GitHub forks?▼

The skill skips forks flagged archived or disabled during listing, and marks forks UNREADABLE when the compare call returns 404. It authenticates through gh api using GITHUB_TOKEN, so access follows the token's permissions.

What happens when GitHub API rate limits are hit?▼

On a 429 response the skill sleeps 60 seconds and retries once; on 5xx errors it waits 10 seconds and retries once. If the initial fork listing fails after retry, the run exits with status FORK_FLEET_API_FAIL and sends an error notification.

How does the skill avoid sending repetitive weekly notifications?▼

It compares the current fork set against a persisted state file and suppresses notifications when there are zero active forks or when nothing crossed the REVIEW tier and every change bucket is empty. Only real deltas trigger a notify message.