Telligence

Self-hosted repo intelligence for maintainers and coding agents.

Telligence indexes repositories and docs, answers questions with citations, and reviews diffs for likely regression risk.

Local-first by default. Cloud AI optional.

Why Telligence exists

Maintainers and coding agents fail when they work without trusted project context.

Telligence provides a private retrieval layer for codebases and docs with:

• grounded answers linked back to source files
• hybrid retrieval for code and documentation
• self-hosted storage and local-first deployment
• agent-friendly workflows for CLI and API clients

Who this is for

• open-source maintainers reviewing incoming changes
• developers onboarding into unfamiliar repositories
• teams that want cited answers instead of vague AI guesses
• coding agents that need trusted repo context

What Telligence is not

The OSS core is intentionally narrow. It is not:

• a generic AI platform
• a compliance suite
• an eDiscovery product
• a broad enterprise memory layer

Core workflows

1. Index a repository or local folder.
2. Ask questions and inspect citations.
3. Review a PR or diff for likely regression risks.
4. Find impacted files, modules, and likely missing tests.

Example prompts:

• Where is authentication handled?
• Summarize the indexing pipeline and cite the key files.
• What modules are likely affected by this config change?
• Review this PR and flag regression risks.

Quick start

git clone https://github.com/shmindmaster/telligence.git
cd telligence
node install.js

Then run:

telligence doctor
telligence onboard

See QUICKSTART.md for the full setup flow.

CLI

The repository build ships these commands:

• telligence doctor
• telligence onboard
• telligence index <path>
• telligence ask "<question>"
• telligence review-pr --base main --head feature-branch

index, ask, and review-pr expect the API to be running locally at http://localhost:8000 unless TELLIGENCE_API_URL is set.

Screenshots

Preview	Description
	Login - Sign in with your account or the demo account (demo@telligence.local).

For the full setup flow, demo credentials, and advanced options, see Quickstart.

---

Core use cases

1. Ground coding agents with direct, real-time access to indexed GitHub repositories and local documentation.
2. Answer developer questions with citations back to source files
3. Run locally by default while keeping the option to use OpenAI or Azure

---

What makes it different

• Built for codebases first - Directly ingest from GitHub repositories, local folders, and uploaded files into one unified retrieval layer.
• Grounded answers - search and chat responses link back to the source material they used
• Flexible deployment - Docker-first, self-hosted, and compatible with local or hosted models
• Agent-ready direction - the project is being packaged around onboarding, MCP, and developer workflows rather than generic "chat with docs"

---

Project status

• supported: onboarding, indexing, search, cited answers, PR review API/CLI
• active development: richer maintainer review UX and agent integrations
• intentionally not part of the OSS hero story: legal/compliance-heavy workflows

See VISION.md for priorities and anti-goals.

Repository layout

• apps/api - API, indexing, retrieval, review workflow
• apps/web - web application
• scripts/installer - onboarding flow
• scripts/setup - setup helpers
• docs/ - architecture, deployment, and contributor docs

Contributing

Start with:

• Contributing
• Code of Conduct
• Security

License

MIT. See LICENSE.