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
35
36
37
38
39

name: parallel-debt-elimination
mode: mapreduce

# Optional setup phase to generate work items
setup:
  - shell: "debtmap analyze . --output debt_items.json"

# Map phase: Process each debt item in parallel
map:
  input: debt_items.json
  json_path: "$.debt_items[*]"
  
  # Commands to execute for each work item (simplified syntax)
  agent_template:
    # Note: Context field is not yet implemented. Pass context via command arguments instead.
    - claude: "/fix-issue ${item.description} --file ${item.location.file} --line ${item.location.line}"

    - shell: "cargo test"
      on_failure:
        claude: "/debug-test ${shell.output}"

  # Parallelization settings
  max_parallel: 10

  # Optional filtering and sorting
  filter: "severity == 'high' || severity == 'critical'"
  sort_by: "priority"

# Reduce phase: Aggregate results (simplified syntax)
reduce:
  - claude: "/summarize-fixes ${map.results}"
    capture_output: true

  - shell: "git merge --no-ff prodigy-agent-*"
    commit_required: true

  # Note: Environment variables in reduce phase are not yet implemented.
  # Pass values as command arguments instead.
  - claude: "/generate-report --fixed ${map.successful} --failed ${map.failed}"