Summary
While reviewing PR #488 I noticed pre-existing drift between README.md and the per-category docs/*.md files. After auditing all 34 categories, the gap is significant enough that I want to check with you on the preferred shape of a fix before opening a 100+-line PR cold.
Drift snapshot
Two kinds of drift exist:
A) Entries in README but missing from docs/*.md — 89 entries across 19 sections. These look like clear drift since docs/ is supposed to be the fuller list. Examples:
finance--crypto.md (18 entries): cnghockey/sats4ai-mcp-server, atxp-dev/atxp, covalenthq/goldrush-mcp-server, …
marketing-sales--crm.md (9 entries): sharozdawa/ai-visibility, brandu-mos/konquest-meta-ads-mcp, …
developer-productivity--utilities.md (9 entries): cafeTechne/antigravity-link-extension, OzorOwn/frostbyte-mcp, …
B) Entries in docs/*.md but missing from README — 1255 entries (after taking the recent tail of each docs file). The README header says "we only show the most recent 30 servers in each category", but actual README per-section counts range from 5 to 75, so the rule isn't strict. Some sections clearly have recent docs additions that never made it to README. Example I noticed first (which led to this audit):
travel--transportation.md: 7 entries in docs, 6 of them at the very bottom (most recent), missing from README. e.g. r-huijts/ns-mcp-server, sunsetcoder/flightradar24-mcp-server, GongRzhe/TRAVEL-PLANNER-MCP-Server.
C) Description mismatches — 11 entries with same slug in both files but different descriptions (e.g. walkojas-boop/regexforge, arcadia-finance/mcp-server, shopsavvy/shopsavvy-mcp-server).
Question
How would you like the cleanup landed?
- One large sync PR — covers all three categories above. Probably ~1300+ lines changed across README + most
docs/ files.
- Section-by-section PRs — one PR per affected category (~19-27 PRs depending on scope), easier to review but heavier on your queue.
- README→docs only — just fix the 89 unambiguous "in README, missing from docs" entries first; revisit the docs→README direction later once you've clarified the truncation rule.
- Close as known debt — you're aware, no action needed.
I can do any of the above. Happy to start with whichever you prefer, or skip if you'd rather handle internally.
For reference, the audit script is just standard re parsing of ^- \[name\](url): description lines per section — happy to share if useful.
— spotted while working on #488 (which is unrelated and self-contained)
Summary
While reviewing PR #488 I noticed pre-existing drift between
README.mdand the per-categorydocs/*.mdfiles. After auditing all 34 categories, the gap is significant enough that I want to check with you on the preferred shape of a fix before opening a 100+-line PR cold.Drift snapshot
Two kinds of drift exist:
A) Entries in README but missing from
docs/*.md— 89 entries across 19 sections. These look like clear drift sincedocs/is supposed to be the fuller list. Examples:finance--crypto.md(18 entries):cnghockey/sats4ai-mcp-server,atxp-dev/atxp,covalenthq/goldrush-mcp-server, …marketing-sales--crm.md(9 entries):sharozdawa/ai-visibility,brandu-mos/konquest-meta-ads-mcp, …developer-productivity--utilities.md(9 entries):cafeTechne/antigravity-link-extension,OzorOwn/frostbyte-mcp, …B) Entries in
docs/*.mdbut missing from README — 1255 entries (after taking the recent tail of each docs file). The README header says "we only show the most recent 30 servers in each category", but actual README per-section counts range from 5 to 75, so the rule isn't strict. Some sections clearly have recent docs additions that never made it to README. Example I noticed first (which led to this audit):travel--transportation.md: 7 entries in docs, 6 of them at the very bottom (most recent), missing from README. e.g.r-huijts/ns-mcp-server,sunsetcoder/flightradar24-mcp-server,GongRzhe/TRAVEL-PLANNER-MCP-Server.C) Description mismatches — 11 entries with same slug in both files but different descriptions (e.g.
walkojas-boop/regexforge,arcadia-finance/mcp-server,shopsavvy/shopsavvy-mcp-server).Question
How would you like the cleanup landed?
docs/files.I can do any of the above. Happy to start with whichever you prefer, or skip if you'd rather handle internally.
For reference, the audit script is just standard
reparsing of^- \[name\](url): descriptionlines per section — happy to share if useful.— spotted while working on #488 (which is unrelated and self-contained)