Cross-Pollinate Hex MCP Servers

---

description: "Audit hex MCP servers for transferable optimizations using diff-driven transfer matrix" allowed-tools: "Bash,Agent,mcp__hex-line__read_file,mcp__hex-line__grep_search,mcp__hex-line__outline,mcp__hex-line__directory_tree,mcp__hex-line__edit_file,mcp__hex-line__write_file"

Cross-Pollinate Hex MCP Servers

Diff-driven audit: find real transferable changes across hex-line-mcp, hex-ssh-mcp, hex-graph-mcp. Each delta is classified, not blindly ported.

Server	Directory	Tools	Role
hex-line-mcp	mcp/hex-line-mcp/	11	Local file ops (source of current deltas)
hex-ssh-mcp	mcp/hex-ssh-mcp/	6	SSH remote ops
hex-graph-mcp	mcp/hex-graph-mcp/	7	Code knowledge graph

---

Phase 0: Change Inventory

Start from actual changes, not theoretical checklists.

git diff --stat -- mcp/hex-line-mcp/
git status -- mcp/hex-line-mcp/

Read each changed file's diff. Classify changes into categories:

Category	Examples
API/schema	New params, changed descriptions, tool registration
Runtime behavior	Edit logic, search logic, error handling
Output normalization	Dedup, truncate, format pipelines
Shared infra	Version sourcing, update-check, coerce, security
Tests	New test coverage
Docs	README, descriptions

Phase 1: Transfer Matrix

For EACH change from Phase 0, check if it applies to hex-ssh and hex-graph.

**Decision categories:**

• APPLY — real gap, same pattern needed in target server
• ALREADY_PRESENT — target already has equivalent implementation
• N/A_BY_DESIGN — change is domain-specific to source (e.g., local-only, SSH-only)
• REJECT — change would hurt target server

**Required evidence:** file + line reference for BOTH source change AND target server status.

Output:

Change	Evidence (hex-line)	hex-ssh status	hex-graph status	Decision	Rationale

Shared optimization checks (always verify these):

Check	What to look for
Dynamic version	createRequire(...)("./package.json") vs hardcoded string in McpServer constructor and checkForUpdates
Missing dependencies	All imports have matching entries in package.json dependencies
Safe process spawning	execFileSync (arg array) vs execSync (shell string interpolation) in production code
CRLF normalization	Consistent .replace(/\r\n/g, "\n") where files are read
Benchmark parity	README claims token efficiency → benchmark.mjs exists with reproducible numbers

Phase 2: Apply Transfers

Implement all `APPLY` decisions from Phase 1.

For each change:

1. Read target file with hex-line tools
2. Apply the change (edit_file or write_file)
3. Verify syntax: npm run check in target package

Phase 3: Local Cleanup (optional)

If Phase 0 revealed drift in hex-line's ow