Every quarter, someone asks: "What features does our product actually have?" The answers vary depending on who you ask. PM lists the marketing features. Engineering lists the technical capabilities. Sales lists what they demo. Nobody has the complete, current picture.
The wiki page titled "Feature List" was last updated 9 months ago by someone who's no longer at the company.
The Wiki Failure Mode
Wikis fail for feature documentation because:
- Features change faster than anyone updates the wiki
- Nobody's job is "maintain the feature wiki"
- Features don't have clean boundaries — they overlap, share code, and depend on each other
- Different audiences need different views (technical vs. user-facing vs. competitive)
The Alternative: Code-Derived Feature Catalogs
What if the feature catalog was automatically generated from the code itself?
Graph-based feature discovery (using dependency clustering) produces a feature catalog that:
- Updates automatically when code changes
- Includes precise file lists, not vague descriptions
- Shows dependencies between features
- Tracks ownership and recent activity
- Is always current because it's derived from the source of truth
A code-derived feature catalog answers:
- "What features do we have?" → clustered feature list with boundaries
- "How complex is each feature?" → file count, dependency count, change frequency
- "Who owns each feature?" → derived from commit history
- "How do features relate?" → cross-feature dependency map
- "What changed recently?" → latest commits per feature
This replaces the stale wiki with a living document that can't go out of date because it's extracted from the code itself.
Originally published on glue.tools. Glue is the pre-code intelligence platform — paste a ticket, get a battle plan.
Top comments (0)