Skip to main content

loci

A personal memory server. You save what you read; loci tags each source with aspects (methodology, knowledge-graph, …), wires up concept edges (citations, wikilinks, co-aspect), and serves the result to Claude Code over MCP so the model can recall the parts of your library that actually matter for what you're writing.

No interpretation graph, no draft pipeline, no PageRank. Just:

save → tag → recall

If you want a deeper read, start at docs/getting-started.md and then docs/architecture.md.

Status

Single-user, local-first. SQLite is the source of truth; raw blobs are content-addressed on disk. Python 3.12+.

Install

# with uv (recommended — isolated environment)
uv tool install loci-wiki

# with pipx
pipx install loci-wiki

The first scan downloads the embedding model (BAAI/bge-small-en-v1.5, ~130 MB) into ~/.loci/models/.

Quick start

# 1. First-run setup: writes ~/.loci/.env (provider keys) and ~/.loci/config.toml
loci config init

# 2. Create a project
loci project create my-research

# 3. Register loci with Claude Code (one-time, user-scope)
claude mcp add loci --transport stdio --scope user -- loci mcp

# 4. Bind a directory to your project (so MCP knows which slug to use)
cd ~/Documents/my-research
loci project bind my-research

# 5. Save sources directly from Claude Code
#    (or via CLI: `loci save https://arxiv.org/abs/1612.03975`)

# 6. Recall in Claude Code
#    `@loci:source://...` for a single resource
#    `loci_recall("how does PPR work")` for ranked chunks with reasons

All user data lives under ~/.loci/. Run loci doctor to see resolved paths.

CLI commands

loci config init                              # write ~/.loci/.env + config.toml
loci doctor                                   # show storage paths + active project
loci server                                   # HTTP API + worker on 127.0.0.1:7077
loci mcp                                      # MCP stdio server (for Claude Code)
loci worker                                   # background worker only

loci project create <slug>                    # interactive wizard
loci project list / info / bind / manage
loci current set <slug>                       # pin project for MCP sessions

loci workspace create / list / add-source / scan / link / unlink
loci scan <project>                           # scan all linked workspaces
loci use [workspace_slugs...]                 # bind workspaces for this session

loci save <url_or_path> [--folder] [--aspects]
loci recall "query" [--aspects ...] [-n 10]
loci aspects [resource_id] [--add ... --remove ... --list-vocab]

loci status [project]
loci export [project]
loci reset                                    # wipe everything

MCP surface (Claude Code)

tool what it does
loci_save ingest URL/file/text, propose folder + aspects via elicitation, write to DB
loci_recall concept-expand + BM25/ANN over chunks with concept-graph rerank
loci_aspects list/edit aspects on a resource (elicitation form for editing)
loci_browse list resources with folder + top aspects, filterable
loci_context project profile + counts + top aspects for the current session
loci_research paper-search sub-agent (deferred to v1.1)

@-mentionable resources:

@loci:source://{resource_id}
@loci:folder://{folder_path}
@loci:aspect://{label}

Source layout

src/loci/
  ui/cli.py              entry point: loci.ui.cli:main
  api/                   FastAPI REST + WebSocket
    routes/              projects, workspaces, sources, aspects, jobs
  mcp/                   MCP server + project resolution
  graph/                 sources, aspects, concept_edges, projects, workspaces
  retrieve/              lex + vec + hyde + concept_expand + pipeline
  capture/               ingest, folder_suggest, aspect_suggest, link_parser
  ingest/                walk → hash → extract → chunk → embed
  jobs/                  queue + worker + handlers (classify_aspects, parse_links, …)
  embed/                 sentence-transformers wrapper
  llm/                   pydantic-ai wrapper
  db/                    schema.sql + connection.py
  config.py              Settings + ~/.loci/ paths

License

MIT.

Metadata

Release files for loci-wiki 0.1.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for loci-wiki 0.1.2
File Size Uploaded Provenance Actions
loci_wiki-0.1.2.tar.gz 1.3 MB Without provenance Details for loci_wiki-0.1.2.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for loci-wiki 0.1.2
File Interpreter ABI Platform Actions
loci_wiki-0.1.2-py3-none-any.whl Python 3 none any Details for loci_wiki-0.1.2-py3-none-any.whl

Total release size: 1.4 MB

Release files / loci_wiki-0.1.2.tar.gz

Download URL loci_wiki-0.1.2.tar.gz
Size 1.3 MB
Tags Source
SHA-256 checksum
How to use checksums
95123db4fd62df10662f0555d277e94919ff68963850fbb52f87aeb89c9a4119
BLAKE2b-256 checksum
How to use checksums
008c8d3a08dbd310db779bc79d73a0b778f8236de18301d9000ebd05989c37c7
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.11.6 {"installer":{"name":"uv","version":"0.11.6","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"macOS","version":null,"id":null,"libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

Release files / loci_wiki-0.1.2-py3-none-any.whl

Download URL loci_wiki-0.1.2-py3-none-any.whl
Size 136.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
c1b51f6c26001da397ddc8fbd886031a1a517e238551263c1d2681cdd50ff33f
BLAKE2b-256 checksum
How to use checksums
e7824dba5a62ae0a95575db3e12f5591144f5871affa3369eccca7e6e71b043f
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.11.6 {"installer":{"name":"uv","version":"0.11.6","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"macOS","version":null,"id":null,"libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

Security for loci-wiki 0.1.2

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page