Tmppr
Pricing
- Free Tier
- Full local features free; cloud runners paid
Summary
Coding agents working in parallel have no shared coordination layer — they overwrite each other's changes, lose context in chat logs, and leave you manually stitching together what ran, what passed, and what's ready to merge. tmppr gives those agents a local forge: PRs, CI, review threads, and merge gates, all running on your machine.
The core workflow is a full pull-request lifecycle that runs on localhost: agents open PRs, trigger CI via your existing GitHub Actions YAML, review each other's diffs inline, and ship only after merge gates clear. An MCP server registers every action as a native tool for Claude Code and Codex, so agents call PR open, review, and merge without glue code. The stacked-PR workflow handles dependent branches in order without manual rebasing. The ceiling appears when teams need shared CI infrastructure — local runners work for solo or small-team setups, but shared minutes require the paid cloud runner option. No API means external systems cannot push events into tmppr without building through the plugin SDK.
Bottom line: Pick tmppr when you want autonomous coding agents to coordinate through structured PRs on a single machine without code leaving your network — plan a different architecture when your CI needs to fan out across multiple machines or integrate with an existing cloud pipeline you cannot replace.
Community Performance Report Card
No community ratings yet. Be the first to rate this tool!
Community Benchmarks Community
Sign in to submit a benchmarkNo community benchmarks yet. Be the first to share a real-world data point.
Pros
Sign in to edit- MCP server auto-registered on init, so Claude Code and Codex agents call PR, CI, review, and merge as native tools — no glue scripts to maintain between your agent loop and the forge.
- Executes existing GitHub Actions YAML locally with live log streaming, which means you reuse CI definitions you already have instead of rewriting them for a new platform.
- Strict 127.0.0.1 binding with no telemetry and no account required for local use, so teams with hard code-leakage policies can run full agent loops without sending source off the machine.
- Full-lifecycle event triggers covering the entire issue-to-merge cycle, which means a failed CI run automatically queues a fix agent rather than leaving the loop waiting for a human to notice.
- Plugin SDK lets you add agent providers, merge rules, and automation triggers from a single installable package without forking core, so your customizations survive upstream updates.
Cons
Sign in to edit- Local CI runners work for a single machine — the moment a team needs shared CI minutes, parallel runners, or a bigger build box, they hit a wall that only the paid cloud runner option unblocks. Teams evaluating cost at that point compare against existing cloud CI they already pay for.
- No public API means external systems — monitoring tools, issue trackers, deployment pipelines — cannot push events into tmppr without a plugin authored against the SDK. Teams that need bidirectional integration with an existing platform end up maintaining a custom plugin as a second system.
- The tool has no listed alternatives in market, but teams whose agent workflows already live inside a hosted platform like GitHub Actions with Copilot or a managed agent orchestration service will find tmppr's local-first model an architectural mismatch rather than a drop-in addition — those teams do not adopt it, they replace it.
Community Reviews
Sign in to write a reviewNo reviews yet. Be the first to share your experience.
About
- Platforms
- macOS, Linux, Windows
- API Available
- No
- Self-Hosted
- Yes
- Last Updated
- 2026-06-18T08:59:05.870Z
Best For
Who it's for
- Teams using Claude Code, Cursor, or similar with agent loops
- Local-first development avoiding code leakage
- Enforcing structured processes on autonomous coding agents
What it does well
- Coordinating multiple coding agents on shared tasks via PRs
- Running local CI and enforcing merge gates for agent-generated code
- Automating review, QA, and fix cycles in agent workflows
Integrations
Discussion Community
Sign in to commentNo discussion yet. Sign in to start the conversation.
Compare Tmppr
Spotted incorrect or missing data? Join our community of contributors.
Sign Up to ContributeCommunity Notes & Tips Community
Sign in to contributeBe the first to contribute. General notes, observations, gotchas, and tips from people who use this tool day-to-day.
Frequently Asked Questions
- Is Tmppr free?
- Tmppr is a paid tool. No permanent free tier is offered.
- Is Tmppr open source?
- No — Tmppr is a closed-source tool. Source code is not publicly available.
- Can I self-host Tmppr?
- Yes. Tmppr supports self-hosting on your own infrastructure.
- What platforms does Tmppr support?
- Tmppr is available on: macOS, Linux, Windows.
Hours Saved & ROI Stories Community
Sign in to contributeBe the first to contribute. Concrete time/cost savings, with context. e.g. "Cut my code review backlog from 4h to 45m per week."
Curated lists that include this category
Most agent-based coding setups devolve into one agent overwriting another’s work, with no record of what was reviewed or why something merged. tmppr addresses this by giving agents a local Git forge — the same pull-request primitives a human team uses, but scriptable end-to-end. Agents grab open issues, open PRs, run CI against your existing GitHub Actions YAML, post inline review comments, and trigger follow-up agents on events like failed CI or requested changes. Merge gates enforce GO/NO-GO from review approval, CI status, and conflict checks before a real `git merge –no-ff` in a detached worktree.
The differentiating feature is the MCP server, auto-registered on `tmppr init`. Every action — PR open, sync, comment, review, rerun CI, merge — is exposed as a native MCP tool, which means Claude Code and Codex call them directly without wrapper scripts. Combined with full-lifecycle event triggers (`pr.created`, `ci.updated`, `review.updated`, and others), the entire issue-to-merge loop can run without a human touching it. You stay in the loop by setting the merge rules; the agents work within them.
The plugin SDK (`@tmppr/plugin-sdk`) lets teams extend the forge without forking: a single package can add agent providers, automation triggers, merge rules, hook executors, event subscribers, and HTTP routes, installed with one command and running on 127.0.0.1 alongside everything else. This matters when your agent provider is not Claude Code or Codex — you wire in a custom command and tmppr treats it as first-class.
The architecture binds strictly to localhost with no telemetry or account requirement for the local tier, which satisfies teams with code-leakage concerns. The constraint is the inverse: teams that need CI to fan out across machines, integrate with an existing cloud pipeline, or expose tmppr as a service to external systems will hit the boundary fast. Cloud burst runners are a paid-only feature, and the absence of a public API means external event sources require a plugin to get data in.