obsei

io.github.obseiPublished by GitHub account obseiv1.1.1Updated Oct 31, 2026

Self-hosted customer feedback with PII stripped at ingest: search, stats and themes with citations.

VerifiedSTDIODesktop onlyKnowledge & MemoryBusiness & Commerce

Overview

AI-generated overview

Self-hosted customer feedback server that redacts personal data at ingest and lets an assistant search feedback, view stats and find themes with citations.

What you can do
  • Search collected customer feedback across channels and languages, with cited evidence for answers.
  • View feedback statistics and stable themes derived from the stored data.
  • Work against a local, self-hosted store whose personal data is redacted before storage and before any LLM sees it.
  • Collect feedback from sources such as app stores and Slack into that store, with optional notification sinks.
When to use it
Useful when you want an assistant to answer questions about your own customer feedback while keeping personal data out of the model, for example for product or support research on self-hosted data.
Requirements
Runs locally as a Python package via uvx (Python 3.11 or newer; version 1.0 or later). Needs an obsei.yaml config, referenced by OBSEI_CONFIG, and a database, optionally set by OBSEI_DB. The database key is supplied through OBSEI_DB_KEY, and the README also mentions a pseudonym salt variable for servers and CI. Optional extras add classifiers, embeddings, source connectors and notifications.
Before you install
The database is encrypted with a key you must supply and keep safe; the README notes that init writes the key and pseudonym salt to a local secrets file, which on servers and CI must be provided as environment variables. Feedback is collected from third-party channels such as app stores and Slack, and optional extras can send notifications to services like Telegram, Discord and email. The README warns that 1.0 is not compatible with the legacy 0.0.x line.

Installation

In SourceWeft

  1. Open obsei in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

[obsei]

obsei

Know what customers are saying. Keep what they said private. The self-hosted, open source alternative to Enterpret and Unwrap that strips PII before any LLM sees it. Collect feedback from every channel and language, redact personal data before it is stored, find stable themes, and let your AI agents answer questions with cited evidence.

Requires Python 3.11 or newer; on older Pythons, pip quietly installs the legacy 0.0.x, so keep the >=1.0 floor. 1.0 is not compatible with 0.0.x; for the legacy toolkit, pip install "obsei<1".

bash
uvx --from "obsei[recommended]>=1.0" obsei init   # or: pip install "obsei[recommended]>=1.0"obsei try                                        # redacted, labelled preview; nothing storedobsei add appstore                               # add sources and sinks without editing YAMLobsei add slackobsei checkobsei run

obsei init writes the database key and pseudonym salt to .obsei/secrets.env (owner-only, git-ignored), which every command loads; on servers and in CI, set OBSEI_DB_KEY and OBSEI_PSEUDONYM_SALT or mount the file.

Optional extras: recommended (local and embeddings), local (in-process classifier), sql, google, apple, names (person-name redaction), embeddings (multilingual themes), notify (Telegram, Discord, email and more via Apprise).

Use it from your AI tool: /plugin install obsei --marketplace obsei/obsei in Claude Code, gemini extensions install https://github.com/obsei/obsei, or uvx --from "obsei>=1.0" obsei mcp in any MCP client. Codex, Copilot, VS Code, Cursor and others: https://docs.obsei.com/guides/ai-tools/

Source: packages/obsei/README.md at commit 392e182

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

4
  1. v1.1.1LatestOct 17, 2026
  2. v1.1.0Oct 16, 2026
  3. v1.0.1Oct 15, 2026
  4. v1.0.0Oct 15, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts