difflore

DiffLore turns your team's past PR/MR review comments into source-backed rules your local AI coding agents can recall before they write code.

The open-source CLI runs locally: import private GitHub or GitLab review history, approve the useful rules, and serve the relevant ones to agents through MCP and hooks. DiffLore never commits, pushes, opens PRs, or posts GitHub comments.

Hosted DiffLore Cloud is optional. The local CLI does not have a hosted PR quota; cloud quotas apply only to managed GitHub App / team workflows.

Why

AI coding agents know public code. They do not know the review decisions your team already made:

• "Use this helper in billing paths."
• "Do not bypass this auth wrapper."
• "This package handles retries; do not add a second loop."
• "This service rejects raw SQL outside migrations."

Those rules are usually buried in old review threads. DiffLore mines them into local memory, keeps source evidence attached, and gives agents the relevant rules before they edit the files that trigger them.

Install

curl -fsSL https://difflore.dev/install.sh | sh

Update later:

difflore update

GitHub import uses your local git remote and GitHub CLI auth:

gh auth login

GitLab import uses a stored PAT with read_api scope:

echo "<TOKEN>" | difflore auth gitlab

Quickstart

Try the demo:

difflore try

Use it in a real repo:

cd your-repo
difflore init
difflore import-reviews --dry-run
difflore import-reviews
difflore memory
difflore memory review
difflore agents install
difflore recall --diff
difflore review --diff all

That flow imports merged review history, turns review comments into local memory candidates, lets you approve or reject them, and wires DiffLore into detected local agents.

Local by default

DiffLore works with private repos and local AI CLIs. A cloud account is not required for the local workflow.

• Local rules and activity are stored in local SQLite.
• difflore import-reviews writes locally unless you explicitly pass --upload.
• difflore cloud login and difflore cloud sync are opt-in.
• Raw local queues are not uploaded by default; cloud sync requires explicit flags for observations, candidates, or telemetry.
• Static exports to AGENTS.md / CLAUDE.md are optional snapshots. Live agents install is the preferred path because it is diff-aware.

Agent support

Run:

difflore agents install
difflore agents status

DiffLore installs MCP entries and lifecycle hooks where the local agent supports them. The installer detects common local coding agents, including Claude Code, Codex, Cursor, Gemini CLI, Windsurf, Goose, OpenCode, Copilot CLI, Crush, Roo Code, Warp, and Antigravity.

Commands

Run difflore --help for the full command list.

Optional Cloud

The cloud layer is for teams that want hosted GitHub App ingestion, shared team rules, dashboards, managed semantic recall, governance, and audit workflows.

difflore cloud login
difflore cloud status
difflore cloud sync
difflore memory team-candidates

Use the local CLI first when you want a no-account path. Use cloud when multiple people need one shared memory and review workflow.

Development

cargo fmt --all --check
cargo check -p difflore-cli
cargo test -p difflore-cli

Issues and PRs are welcome. Do not include secrets, private PR text, or private code in examples.

For suspected vulnerabilities, email hello@difflore.dev instead of opening a public issue. See SECURITY.md.

License

Apache 2.0. See LICENSE.