Follows every commit
A signature-verified webhook notifies Repotrix on each push, and it reads exactly what changed. Nothing waits on a manual trigger or a scheduled job.
Most tools that chat with your repository answer from a snapshot taken at setup. The moment your team merges, that snapshot is out of date. Repotrix works differently: it follows every commit, understands what changed, and keeps your documentation, diagrams, and answers in step with the code. Available on the web and as a native VS Code extension.
When documentation falls out of sync, people quietly stop trusting it. Repotrix is built so that does not happen. Here is what makes it work.
A signature-verified webhook notifies Repotrix on each push, and it reads exactly what changed. Nothing waits on a manual trigger or a scheduled job.
Documentation, diagrams, security findings, and the search index update against the new commit together, so they never disagree with each other or with the code.
Each change is reconciled against the dependency graph, so you can see precisely where the system has moved away from its documentation.
The longer Repotrix follows a repository, the better it understands your patterns, so its reviews and answers grow more precise the longer you use it.
The difference isn't the feature list — it's freshness. Here is the same job, side by side.
A guided walkthrough of the real product, from the first ingestion to a merge that brings everything back in sync. Select any chapter to begin.
Connect any GitHub repo. Ingested in minutes.
Settles card and UPI payments end to end. Owns webhook dispatch, idempotency keys, and the reconciliation cron…
Read full documentation →Documentation, diagrams, security, chat, and reviews all draw on the same dependency graph and search index, and they refresh together as the code changes. Nothing contradicts anything else, on the web or in VS Code.
The engine behind everything else. A signature-verified webhook reports each push, the worker reads precisely what changed, and your documentation, diagrams, security posture, and search index update on their own. Nothing waits on a manual step or a scheduled job.
Technical documentation generated from your code and rewritten whenever a merge changes what it describes, so it reflects what actually shipped.
Architecture, dependency, and sequence diagrams in Mermaid that regenerate on each push, so the picture always matches the current system.
Graph-aware retrieval that searches the latest commit and cites the exact files behind every answer, so you can verify what it tells you.
Every pull request is reviewed against your current documentation, and the findings are posted back to GitHub for the team to act on.
Exposed secrets, PII, and licensing issues are re-evaluated as the code changes, with a health score that tracks each repository over time.
Role-aware onboarding paths built from the live codebase, so new engineers learn the system as it stands today.
When a change pulls the code away from its documentation, Repotrix posts a short summary to Slack or Teams so the team stays aware.
Redaction events, webhook rotations, and token changes are recorded in a tamper-evident log you can export when it is needed.
The same up-to-date documentation, diagrams, security, and chat, without leaving the editor. Install once, and Repotrix lives in your sidebar, answering from the latest commit while you work.
Strong security practices overall — robust password hashing, comprehensive log redaction and signature-verified webhooks. Watch the default CORS config and JWTs passed in WebSocket URLs.
CORS_ORIGINS allows any origin (*); restrict it in production.Ask about the file you are in and get answers grounded in the current commit, without leaving the editor.
Open up-to-date documentation and architecture diagrams beside your code, drawn from the same context as the web app.
Read health scores, security findings and pull-request feedback in the place where you write code.
Walk a new engineer through the codebase from a native sidebar, from their first clone to their first commit.
Connect your AI assistant to Repotrix over the Model Context Protocol - Claude Code, Cursor, Copilot, Gemini, ChatGPT and other MCP-compatible tools. Whatever tool you code in can ask grounded, always-fresh questions about your repositories - no copy-pasting, no stale context. Available on paid plans.

list_reposask_codebasesearch_codeget_architecturewhats_changedSecrets are encrypted, requests are rate-limited, responses are cached against the live commit, and every token is metered. It is ready for a real team from the first day.
Webhook secrets and GitHub PATs are encrypted with Fernet — keys live in the environment, never in the database.
Gemini responses are cached against the latest commit hash, so a new push invalidates them naturally — fresh answers, small bill.
Fixed-window rate limiting with dedicated budgets per feature. Noisy workspaces can't starve the rest.
Per-model counters track tokens-in and tokens-out for every RAG turn, with full usage visibility in your dashboard.
Add a fine-grained token and a single webhook. Credentials are encrypted at rest and never shown back to you.
A background worker clones the repository, embeds it into Pinecone, and builds the dependency graph that grounds every answer.
From the next push onward, your documentation, diagrams, security, and answers update on their own. There is nothing to regenerate by hand.
Six perspectives on what changes when your codebase knowledge stays current.
Half of my code-review work just disappeared. The PR companion catches the obvious bugs and lint issues so I can focus on logic and design.
Our architecture diagram used to be out of date within a month. Now it redraws itself on every merge, and I trust it enough to present in a stand-up without screenshots and arrows.
I onboarded onto a five-year-old codebase in two days instead of two weeks. The chat answered everything I would have spent an afternoon grepping for — and it was actually current.
Once the docs stopped going stale, people started trusting them again. That alone changed how the team works, and ramp time dropped from weeks to days.
The RAG is genuinely sharp, and the answers track the code as it changes. It answers faster than I can switch tabs, and the citations land on real files I can verify.
Setup took ten minutes. Three months later it's running on every service we own and we haven't touched the configuration once — it just stays current.
Connect a repository in a few minutes. From then on, every merge keeps your documentation, diagrams, and answers current, on the web and in VS Code, with nothing left for you to maintain.