From programmator
Install and configure programmator — autonomous coding agent orchestrator. Use when the user asks to install programmator, set up configuration, or get started.
npx claudepluginhub alexander-akhmetov/programmator --plugin programmatorThis skill is limited to using the following tools:
Help the user install and configure programmator.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Searches prompts.chat for AI prompt templates by keyword or category, retrieves by ID with variable handling, and improves prompts via AI. Use for discovering or enhancing prompts.
Checks Next.js compilation errors using a running Turbopack dev server after code edits. Fixes actionable issues before reporting complete. Replaces `next build`.
Help the user install and configure programmator.
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).
programmator --helpclaude --versionpi-coding-agent --versionopencode --versioncodex --version
If none are installed, tell the user they need at least one.~/.config/programmator/config.yaml (global) or .programmator/config.yaml (per-project override).programmator start ./plan.md
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):
~/.config/programmator/config.yaml (global).programmator/config.yaml (per-project)Run programmator config show to see all resolved values and their sources.
executor: pi
pi:
provider: anthropic # or "openai"
model: sonnet # or "gpt-4o"
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
Code with pi, review with Claude Opus:
executor: pi
pi:
provider: anthropic
model: sonnet
review:
executor:
name: claude
claude:
flags: "--model opus"
git:
auto_commit: true
Or via CLI: programmator start ./plan.md --auto-commit
| 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 |
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
programmator start <thing> auto-detects the source type:
./plan.md) → plan filepro-1a2b) → ticket ID (requires ticket CLI)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