prodigy 0.4.4

Turn ad-hoc Claude sessions into reproducible development pipelines with parallel AI agents
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34

# Example: MapReduce workflow using command output as input
# This demonstrates using shell commands to generate work items
name: analyze-rust-files
mode: mapreduce

# Setup phase - runs once before map phase
setup:
  - shell: "echo 'Starting analysis of Rust files...'"
  - shell: "pwd | xargs -I {} echo 'Repository: {}'"

# Map phase - processes items in parallel
map:
  # Use find command to generate list of Rust files
  input: "find . -name '*.rs' -type f | head -20"

  # Template for each parallel agent
  agent_template:
    - shell: echo "Analyzing ${item}"
    - shell: |
        wc -l "${item}" | awk '{print "Lines in " $2 ": " $1}'

  # Run up to 5 agents in parallel
  max_parallel: 5

  # Timeout configuration (optional)
  # Note: agent_timeout_secs can also be configured via timeout_config block
  # See docs/examples.md Example 8 for full timeout_config syntax
  timeout_config:
    agent_timeout_secs: 30  # Maximum time per agent (seconds)

# Reduce phase - runs once after all map agents complete
reduce:
  - shell: echo "Analysis complete. Results available in map.results"
  - shell: echo "Total items processed"