ContextLane

Open-source context ingestion pipeline for persistent AI agents.

MemoryLane gives agents memory. ContextLane gives agents something worth remembering.

Turn files, folders, URLs, repos, and notes into structured context with citations — then optionally sync to MemoryLane for persistent agent recall.

Why ContextLane?

Agents fail when context is a pastebin. ContextLane standardizes:

source → load → normalize → chunk → extract → cite → save → optional MemoryLane sync

Every ingestion run can produce:

• Summary — high-level overview
• Key facts — statements with confidence
• Decisions — active / historical
• Action items — TODOs with priority
• Entities — people, tools, URLs
• Tags — auto topics
• Citations — line-level source refs
• MemoryLane records — ready to sync

Local usage is free and open-source. Cloud mode uses Talocode when configured.

Install

pip install -U contextlane
contextlane demo

Python 3.10+.

From source:

git clone https://github.com/talocode/contextlane.git
cd contextlane/python && pip install -e .

Supported sources

CLI

contextlane --help
contextlane --version
contextlane init                    # ~/.contextlane/
contextlane doctor
contextlane demo
contextlane ingest <path>
contextlane ingest-url <url>
contextlane ingest-github <repo>
contextlane runs list
contextlane runs show <runId>
contextlane search <query>
contextlane recall <query>
contextlane sync memorylane <runId>
contextlane export <runId> -o file.json
contextlane import file.json
contextlane serve --port 3060

Auto-sync during ingest:

contextlane ingest ./README.md --sync

Python SDK

from contextlane import ContextLaneClient, ingest, search

# Local engine (no network)
run = ingest("./README.md")
print(run["id"], len(run.get("extraction", {}).get("facts", [])))

results = search("authentication")
for r in results:
    print(r.get("run_id"), (r.get("chunk") or {}).get("text", "")[:120])

# Remote HTTP API
client = ContextLaneClient("http://localhost:3060")
print(client.health())
print(client.list_runs())

HTTP API (local server)

contextlane serve --port 3060

curl -X POST http://localhost:3060/v1/contextlane/ingest \
  -H 'Content-Type: application/json' \
  -d '{"input":"./README.md"}'

curl -X POST http://localhost:3060/v1/contextlane/search \
  -H 'Content-Type: application/json' \
  -d '{"query":"What is this?"}'

MemoryLane sync

Priority:

1. HTTP — CONTEXTLANE_MEMORYLANE_URL
2. CLI — memorylane on PATH
3. Fallback — ~/.contextlane/runs/<runId>/memory-records.json

contextlane sync memorylane <runId>

Cloud mode

export CONTEXTLANE_CLOUD_MODE=true
export TALOCODE_API_KEY=your_key
# optional: TALOCODE_BASE_URL=https://api.talocode.site

Related packages

Talocode ecosystem

More: github.com/talocode · talocode.site

Links

• PyPI: https://pypi.org/project/contextlane/
• npm: https://www.npmjs.com/package/@talocode/contextlane
• GitHub: https://github.com/talocode/contextlane

License

MIT © Talocode