Thanks for using Exegete (formerly qualcoder-mcp). This is an experimental alpha built by one researcher. Feedback, bug reports, and feature ideas are genuinely wanted: they directly shape what gets built next.

Everything goes through GitHub Issues:

• Bug reports: please include what you did (the tool calls or the conversation step), what you expected, and what happened instead. Never paste sensitive research data into an issue; a synthetic or redacted example is perfect. (For what happens to research data when you USE the tool, what leaves your machine and what stays local, see PRIVACY.md.)
• Questions: check the INSTALL.md troubleshooting section and AI_CODING_WORKFLOW.md first, then open an issue. Questions are welcome; yours is probably the next person's too.
• Feature requests and ideas: very welcome. The release philosophy is to develop the next capabilities together with early users, so "it would help my analysis if…" issues are exactly what's wanted.

Issues are public and searchable: every answered question helps the next researcher who hits the same thing, and others can confirm a bug or add detail. That's why there is one channel, and why it isn't email.

The author's email address in the package metadata (pyproject.toml) is an authorship signature: it identifies who wrote this software. It is not a support channel, and support requests sent by email will not receive a reply. This isn't unfriendliness: answers buried in a private inbox help exactly one person once, while answers on GitHub Issues help everyone, permanently. Please use GitHub Issues instead.

• This is alpha software: always work on copies of your projects (the copy_project_to_workspace tool exists for exactly this), and keep your own backups of important data.
• Remember the data flow: everything a tool returns, interview text included, enters your AI conversation and is sent to whichever AI provider your host uses (Anthropic for Claude hosts; no external provider at all with a fully local host). Prefer synthetic or consented data, and check your ethics/GDPR position before using real participant data; see PRIVACY.md.
• If you also use QualCoder, close the project there before any writing. Writes are refused while the project's lock file is active (QualCoder 3.8.2 has the project open, or another conversation using Exegete is writing to it). QualCoder 4.0 writes no lock file, so there Exegete can only report signs that the project may be open, which can miss an open window; a write into a project open in any QualCoder window can lose work.
• Include in bug reports: your QualCoder version, if you use QualCoder (a 3.8.x release or 4.0; see "Supported QualCoder versions" in TOOLS.md for the supported project schemas), the project's schema version (the databaseversion value in the schema block that get_current_project returns), the Exegete version (exegete --version in the environment you installed into, or python -m exegete.server --version for a git install; pip show exegete, pipx list and uv tool list still work and spell 0.15.1-alpha as 0.15.1a0; an install made under the earlier name answers qualcoder-mcp --version too), your MCP host (Claude Desktop, Claude Code, LM Studio, other), and the toolset (EXEGETE_TOOLSET: core, or full when the variable is not set; lifecycle is the earlier name of full, before 0.15.0). With the Claude Desktop extension, the version is in the name of the .mcpb file you installed, and the toolset is core when its switch "Limit Exegete to its small tool set" is on, and full otherwise. The bug report template asks for all of these.