Programmator setup

Programmator Setup

Help the user install and configure programmator.

Installation

Requires Go 1.26+.

go install github.com/alexander-akhmetov/programmator/cmd/programmator@latest

Or download a binary from GitHub Releases for linux/darwin (amd64/arm64).

Setup steps

1. Install the binary (see above). Verify: programmator --help
2. Check that an executor is available — programmator needs a coding agent to do the work:
   • Claude Code (default): claude --version
   • pi coding agent: pi-coding-agent --version
   • OpenCode: opencode --version
   • Codex: codex --version If none are installed, tell the user they need at least one.
3. Optionally create a config — programmator works with zero config (defaults to Claude Code as executor). Only needed if the user wants to use a different executor, customize review agents, or enable auto-commit. Config goes in ~/.config/programmator/config.yaml (global) or .programmator/config.yaml (per-project override).
4. Try it out — create a simple plan file and run it:

programmator start ./plan.md

Configuration

Programmator works with zero config. It defaults to Claude Code as executor with 9 parallel review agents. Only create a config if you need to customize something.

Config file locations (highest priority last):

1. Embedded defaults (built into binary)
2. ~/.config/programmator/config.yaml (global)
3. .programmator/config.yaml (per-project)
4. CLI flags

Run programmator config show to see all resolved values and their sources.

Use pi coding agent instead of Claude Code

executor: pi
pi:
  provider: anthropic   # or "openai"
  model: sonnet          # or "gpt-4o"

Custom review agents

Default: 9 agents run in parallel (bug-shallow, bug-deep, architect, simplification, silent-failures, claudemd, type-design, comments, tests-and-linters). You can replace them:

review:
  agents:
    - name: code-review
      focus:
        - correctness
        - error handling
        - test coverage

Or point to a full prompt file:

review:
  agents:
    - name: code-review
      prompt_file: ".programmator/prompts/review/code-review.md"

Or select a subset of defaults:

review:
  include:
    - bug-deep
    - architect
    - tests-and-linters

Or exclude specific agents:

review:
  exclude:
    - comments
    - type-design

Use a different executor for review

Code with pi, review with Claude Opus:

executor: pi
pi:
  provider: anthropic
  model: sonnet

review:
  executor:
    name: claude
    claude:
      flags: "--model opus"

Auto-commit after each task

git:
  auto_commit: true

Or via CLI: programmator start ./plan.md --auto-commit

Config keys

Key	Default	Description
max_iterations	50	Maximum loop iterations before forced exit
stagnation_limit	3	Exit after N consecutive iterations with no file changes
timeout	900	Seconds per executor invocation
executor	claude	Which coding agent to use ("claude", "pi", "opencode", or "codex")
claude.flags	""	Additional flags passed to the claude command
claude.config_dir	""	Custom Claude config directory
claude.anthropic_api_key	""	Anthropic API key (overrides env)
pi.flags	""	Additional flags passed to pi-coding-agent
pi.config_dir	""	Custom PI_CODING_AGENT_DIR
pi.provider	""	LLM provider for pi ("anthropic", "openai")
pi.model	""	Model name for pi ("sonnet", "gpt-4o")
pi.api_key	""	API key for the configured pi provider
opencode.flags	""	Additional flags passed to opencode
opencode.config_dir	""	Custom OPENCODE_CONFIG_DIR
opencode.model	""	Model in "provider/model" format
opencode.api_key	""	API key for the configured provider
codex.flags	""	Additional flags passed to codex
codex.model	""	Model name (e.g. "o3", "gpt-5-codex")
codex.api_key	""	OpenAI API key
ticket_command	tk	Binary name for the ticket CLI
git.auto_commit	false	Auto-commit after each phase completion
git.move_completed_plans	false	Move completed plans to a completed/ directory
git.completed_plans_dir	""	Directory for completed plans (default: plans/completed)
git.branch_prefix	""	Prefix for auto-created branches (default: programmator/)
review.max_iterations	3	Maximum review fix iterations
review.parallel	true	Run review agents in parallel
review.include	[]	Subset of built-in review agents
review.exclude	[]	Remove specific default review agents
review.agents	[]	Custom review agents (replaces defaults when non-empty)
review.validators.issue	true	Cross-agent false-positive validator
review.validators.simplification	true	Simplification value validator

Plan file format

Minimal plan:

# Plan: Fix bugs

## Tasks
- [ ] Fix the off-by-one error in loop.go
- [ ] Add nil check in handler.go

With validation commands (run after each task):

# Plan: Add authentication

## Validation Commands
- `go test ./...`
- `make lint`

## Tasks
- [ ] Add JWT middleware
- [ ] Add login endpoint
- [ ] Add tests for auth flow

Source detection

programmator start <thing> auto-detects the source type:

• File paths (e.g. ./plan.md) → plan file
• Everything else (e.g. pro-1a2b) → ticket ID (requires ticket CLI)

Commands

programmator start ./plan.md              # execute a plan
programmator start ./plan.md --auto-commit # with git workflow (branch + commits)
programmator start pro-1a2b               # execute a ticket
programmator review                       # review current branch vs main
programmator review --base develop        # review against a different base
programmator run "explain this codebase"  # one-off prompt to the coding agent
programmator config show                  # show resolved config