Line-level reasoning with receipts
Pruve indexes repository history and links a line of code to the issue, pull request, review comment, benchmark, or commit that explains it. The evidence is openable rather than summarized away.
Pruve
Pruve is a read-only developer tool that explains why code exists by linking lines to repository history and other recorded evidence. It helps engineers and agents inspect the reasoning behind code changes without guessing when the context was never written down.
What Pruve does
Pruve is a developer tool that reads a repository’s history and explains why a line of code is the way it is, with linked receipts you can open. If no evidence exists, it says so instead of inventing a reason.
It is built for codebases where the original discussion lives in issues, pull requests, review comments, or agent sessions rather than in the source file itself. The product is read-only for repository access, and it is also exposed through MCP so agents can ask for reasoning before they edit code.
Features
Coverage and explanation chain
The web app highlights whether a line can account for itself, and the app can show the full chain behind a change in order: issue, discussion, decision, and commit.
Search across recorded reasoning
Instead of a generic code search, Pruve searches recorded reasoning such as commit messages, pull request threads, review comments, and agent sessions.
Read-only agent access over MCP
Pruve exposes read-only MCP tools so agents can ask why a line exists before they edit code. The product says this keeps reasoning available to agents without giving them write access.
Optional documentation gate
A pull-request documentation gate can optionally fail changes that do not have recorded reasoning attached. The gate is advisory by default.
Organisation-based pricing model
Pricing excludes lookups with receipts and bills only generated answers, using a pooled allowance rather than per-person seats.
Common use cases
Investigate a suspicious line of code
Use Pruve when a developer or reviewer finds an odd constant, a fragile conditional, or a merge-sensitive line and needs the original reasoning before making a change.
Onboard into code with history attached
Use it when a new team member needs the decision trail behind a subsystem, not just the current implementation, so they can understand why the code looks the way it does.
Trace the full decision chain
Use it to reconstruct a change from issue to benchmark to review to commit when the context was scattered across pull requests and comments instead of a single document.
Let agents ask before they modify code
Use the MCP tools when an agent is about to edit unfamiliar code and should check the receipts first rather than guessing at intent.
Encourage documented changes in review
Use the optional documentation gate when you want pull requests to carry recorded reasoning and you are willing to fail or flag changes that do not.
Pros and Cons
Pros
- Links code to supporting evidence instead of offering an ungrounded guess.
- Keeps lookup-style receipt access free, including open source usage.
- Works read-only against GitHub or GitLab organisations, avoiding repository writes.
- Provides a path for agents to ask before modifying files through read-only MCP tools.
- Can optionally enforce documentation habits with a pull-request gate.
Cons
- The hosted product is not self-hosted; self-managed GitLab and GitHub Enterprise Server are only mentioned for Enterprise sources.
- Generated answers are metered after the included monthly allowance, so heavy use of model-written summaries can add cost.
- The product only explains what it can support with evidence; if the reasoning was never recorded, it will not infer a likely answer.
FAQ
Does Pruve write to my repository?
No. Pruve reads repository history and related evidence in read-only mode. The only write permission it can be granted is an optional pull-request check run if you enable the documentation gate.
What happens if there is no evidence for a line?
It says so and stops. When there is no recorded reasoning, Pruve reports what it searched and how much of it it found instead of guessing.
What is counted against usage?
Generated summaries are metered. The pricing page says the first 2,000 generated answers a month are included for a private organisation plan, and further generated answers are billed after that.
Is open source free?
Open source is free forever. The product page says public repositories are covered for free and receipts remain free in every plan.
Can Pruve be self-hosted?
Not yet for the hosted product. Enterprise covers self-managed GitLab and GitHub Enterprise Server as sources, while the platform itself is hosted.
Quick Facts
- Category
- Developer Tool
- Source domain
- pruve.dev
- Primary surfaces
- Web app, GitHub direct view via extension soon, MCP for agents
- Repository access
- Read-only
- Pricing model
- $99 per organisation per month for private orgs, with generated answers pooled and open source free
- Platform support
- GitHub and GitLab organisations; Enterprise mentions self-managed GitLab and GitHub Enterprise Server as sources
Alternatives à Pruve
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
hob
hob is an independent workspace for coding agents that keeps agent sessions, terminals, history, and follow-up work organized around the tools and providers you already use. It is aimed at developers who want local control over routing, history, and workspace structure rather than a bundled model stack.
Manta AI
Manta AI is an autonomous web app testing tool for teams that want to map application behavior, catch regressions, and generate tests without writing scripts or maintaining selectors. It works from a URL and supports plain-English test flows, run results with screenshots, and scheduled or deployment-triggered checks.
SonOf
SonOf connects to your repo and PM tool, audits the codebase and surrounding product context, and turns approved work into shipped tickets with senior engineering review. It is aimed at founders and engineering leaders who need backlog help without hiring a full team immediately.
Ghost
Ghost est un assistant IA en terminal pour discuter, générer du code et lancer des tâches en ligne de commande. Modèles gratuits, Linux, macOS, Windows, open source.