Is your team facing the same problem:
The old man had a chart in his head, but the new man had to go through the source。
Each code review has been conducted with varying understandings and very inefficient discussions。
Today's presentation of graphidy, an open source knowledge mapping tool, is suitable for a technical team that needs to share structured knowledge with a team and not for small individual projects。
From personal drawings to team sharing
Graphify's design is clear: the drawings should be submitted to git, so that each member of the team can start with a map。
When a person runs a command to generate a spectrograph and submit it, the other person pulls the replacement code and uses it directly。
It even embeds git merge diver, and when both developers submit a map update at the same time, it automatically takes place and collects and does not appear as a conflict mark。
Waterline structure design
Graphify's core is a clear seven-stage line:
Found a file poignant to extract node and edge build poignant poignant community detection analysis generating report export。
Each phase is a stand-alone module, which communicates through a pure python dictionary and networkx diagram, with no shared status。
This design allows it to be used both as an ai code assistant skill and as an independent repository to integrate into your ci/cd process。
Mcp server: let ai directly query the map
It's one of the toughest nuclear functions in the country。
It exposes the map to the mcp server and provides structured interfaces such as query graph, get node, get neigbors, shortest path。
Supports both the stdio and http modes of transmission where the team can deploy a shared process where all developers ide point directly to the same address。
The more advanced use is to combine pr impact analysis: running a graphify prs 42 allows an in-depth look at the extent of the impact of a pr on the map and early detection of potential combined conflict risks。
Security and privacy
For enterprise-level users, privacy is the central consideration。
Graham's code files are all processed locally through the tree-sitter without sending anything to an external server。
Pure code items can even run completely offline without any api key。
It does not have telemetry, does not use tracking and does not analyse reports。
Team deployment orders
# install
uv tool install greyy
# register skills and install git hook
graphify install
graphify hook install
# generate a map
i don't know.The generated graphy-out/ directory is available when submitted to git。
Transferred to the head of the technical team and the architect to address the problem of inconsistent team code awareness。
Focus on us, next time the graphidy in the big moActual performance in norepo。
This post is part of our special coverage global voices 2011。
