Use the right agent inside one shared process.

ProPR makes the agent a per-task choice. Use one model for broad context, another for deep implementation, and a faster model for review. Bring existing provider subscriptions or API keys; planning, implementation, and review still share one process.

One process, multiple brains PR path intact

Task / PlanScope, context, branch

ClaudeDeep reasoning

CodexFast implementation

OpenCodeFlexible routing

One GitHub PR & LogReview, cost, follow-ups

Pick the agent per phase. ProPR keeps planning, execution, cost, and GitHub review in one record.

Best for complex architectural changes, deep refactors, and high-reasoning planning phases.

Fast, deeply integrated implementation using OpenAI's flagship coding models.

Google's multi-model CLI agent: Gemini, Claude, and GPT-OSS backing models through one container and credential mount.

Highly cost-efficient open-weight models for standard reviews and implementation passes.

Provider-agnostic open-source agent with a lineup of free models — zero-cost passes and comparison baselines.

Switch agents mid-task and keep one audit trail.

A task can start with one agent for broad context, move to Claude for a tricky implementation, and use Codex for a follow-up fix. No matter which agent runs, ProPR leaves one coherent audit trail: what was planned, which agent ran, exactly what it cost, and what landed in GitHub.

ProPR bundles its Operator Agent Skill for Codex, Claude, Antigravity, and OpenCode. The installed skill teaches the assistant to delegate through GitHub and ProPR after a human creates or triages the issue. That human controls repository, provider, system, membership, and access permissions; the assistant never grants or broadens them. The CLI is an optional guided path that safely installs, checks, updates, or removes the skill.

Phase handoff One task

Plan Claude Fable

Implement GPT 5.6 Sol

Review Claude Opus

Fix GPT 5.6 Sol

Route the work by phase while the task history, cost record, branch, and pull request stay connected.

Switch by phase

Pick the agent and model that fits each phase: planning, context analysis, implementation, review, follow-ups, and repository suggestions.

Compare agents on the same task

Send one request to several agent and model combinations. Compare their pull requests side by side, then merge the strongest result.

Track the cost

Usage logs show which model handled the task, how long it took, and what it cost. Agent Tank adds subscription and rate-limit visibility across supported agents.

The focused ProPR agent configuration panel showing enabled Claude and Codex models, context limits, IDs, aliases, and defaults. — Agent configuration exposes the exact model IDs, readable aliases, context limits, and defaults.

The ProPR playground model picker with supported agents and models selectable for side-by-side runs. — The playground sends one request through several selected models.

The focused ProPR usage log columns showing model, tokens, cost, and duration. — The usage log makes the cost of each model call concrete.

One issue. Two agent runs. Two pull requests.

Issue #153 asked for abuse-prevention controls. ProPR ran it through Codex with GPT-5.5 and Claude Code with Claude Fable 5. Each run returned its own branch, implementation summary, and pull request for direct comparison.

GitHub pull request 168 from issue 153, showing the GPT-5.5 implementation summary for abuse-prevention controls. — **Codex**GPT-5.5integry/digvin #168

GitHub pull request 169 from issue 153, showing the Claude Fable 5 implementation summary for abuse-prevention controls. — **Claude Code**Claude Fable 5integry/digvin #169

Bring a chat client to the agent workflow.

The MCP integration lets a chat client request work from your ProPR instance. Your configured coding agents still execute on your infrastructure. Direct access and Connect use the same core tools, resources, and prompts.

ProPR exchanges text and structured data. Speech input and output belong to the host, whose voice mode may not expose MCP tools. Browser authentication handles secrets and consent.

See how ProPR fits into the landscape.

Compare the shared PR workflow with hosted agent UIs, AI IDEs, and direct CLI tools.

Compare ProPR against the alternatives →