Problem
Each ./mxcli -p app.mpr -c "COMMAND" invocation opens a new connection to the MPR file. When exploring a project, AI agents typically need 10-20 sequential commands, each paying the connection overhead (Connected to: project.mpr).
Proposal
Option A: Semicolon-separated commands
./mxcli -p app.mpr -c "DESCRIBE MICROFLOW A.X; DESCRIBE MICROFLOW A.Y; DESCRIBE ENTITY A.Z"
Option B: Stdin/pipe mode
echo "DESCRIBE MICROFLOW A.X
DESCRIBE MICROFLOW A.Y
DESCRIBE ENTITY A.Z" | ./mxcli -p app.mpr --stdin
Or from a file:
./mxcli -p app.mpr < commands.txt
Option C: Both
Support both semicolons in -c and stdin pipe mode.
Benefits
- Eliminates N-1 connection overhead for N commands
- Enables AI agents to batch all exploration queries into one call
- Reduces wall-clock time for project exploration significantly
Priority
High — directly reduces the number of round-trips, which is the main bottleneck for AI-assisted exploration.
Problem
Each
./mxcli -p app.mpr -c "COMMAND"invocation opens a new connection to the MPR file. When exploring a project, AI agents typically need 10-20 sequential commands, each paying the connection overhead (Connected to: project.mpr).Proposal
Option A: Semicolon-separated commands
./mxcli -p app.mpr -c "DESCRIBE MICROFLOW A.X; DESCRIBE MICROFLOW A.Y; DESCRIBE ENTITY A.Z"Option B: Stdin/pipe mode
Or from a file:
./mxcli -p app.mpr < commands.txtOption C: Both
Support both semicolons in
-cand stdin pipe mode.Benefits
Priority
High — directly reduces the number of round-trips, which is the main bottleneck for AI-assisted exploration.