upstream-research

Generates architecture, changelog, and sync documentation for upstream project directories.

1|Updated Feb 2, 2026
One-click install
npx skills add https://github.com/rockcookies/skills --skill upstream-research-rockcookies
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: upstream-research
Source: https://github.com/rockcookies/skills/tree/main/.agents/skills/upstream-research
Command: npx skills add https://github.com/rockcookies/skills --skill upstream-research-rockcookies

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Keeping documentation for vendored or mirrored upstream projects current is tedious: you must detect whether the upstream changed, diff commits, and rewrite architecture notes by hand. This Skill automates that research loop and produces four consistent documents per project. ## Core Features & Use Cases - Mode-gated sync: Compares the upstream git HEAD against the recorded last_synced_hash to choose FULL, DIFF, FULL_NO_GIT, or SKIP modes, with rollback warnings and --force / --no-git-check overrides. - Four-file documentation output: Writes architecture.md, changelog.md, sync.json, and extend.md under docs/upstream/<project>/, with incremental changelog prepending for DIFF runs. - User-controlled rules: An extend.md file with protected USER_RULES and auto-regenerated AI_SUGGESTIONS sections lets users steer focus areas without losing customizations. - Use Case: After pulling new commits into upstream/some-lib, run the research flow to get an updated changelog of interface changes and a refreshed architecture overview without reading every diff manually. ## Quick Start Ask the assistant to research the upstream project named my-lib and update its documentation under docs/upstream.

Frequently Asked Questions about upstream-research

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

FAQPage Schema
How do I document changes in an upstream repository automatically?▼

Run the research flow on the upstream/<project> directory. It compares the current git HEAD against the last synced hash, analyzes the commit log and diff, and writes an updated changelog.md plus a refreshed architecture.md under docs/upstream/<project>/.

How do I force regeneration of upstream documentation?▼

Pass the --force flag when invoking the research flow. This rebuilds all four documents even when the upstream HEAD matches the recorded last_synced_hash, which would otherwise result in a SKIP.

Can I analyze an upstream directory that is not a git repository?▼

Yes. Use --no-git-check, or let the skill detect the missing git metadata automatically. It switches to FULL_NO_GIT mode, performs pure directory analysis, and sets hash-related fields in sync.json to null.

What happens when the upstream diff is very large?▼

When more than 200 files changed, the skill stops reading individual diffs and relies only on git diff --stat output. The changelog notes that the update was a large-scale change shown as statistics only.

Will my custom rules in extend.md be overwritten?▼

No. The USER_RULES section is read-only during research; only the AI_SUGGESTIONS block between its markers is regenerated. If the USER_RULES markers are damaged, the skill warns and refuses to overwrite the file.