Releases: codician-team/growmos
Releases · codician-team/growmos
Release list
growmos 0.1.5
Discoverability release — growmos is now listed on the official MCP Registry as com.codician/growmos (DNS-verified Codician namespace) and installable from PyPI.
Changes since 0.1.3:
- Truly hands-off in Claude Code:
SessionStarthook injects the graph brief and pending work;Stophook keeps the agent going until the graph is up to date (loop-safe). Protocol block tells other CLIs to run the loop unasked. - Extraction cap is a speed bump, not a wall:
growmos next --force,growmos config max_docs_per_run 0, default 50/day. Newgrowmos configcommand. growmos integrate mcpwrites.mcp.json/.cursor/mcp.json; README section on registering the MCP server in any client.server.json+mcp-name: com.codician/growmosmarker for the MCP Registry; publishing guide indocs/mcp-registry.md.- Docs: configuration reference (all keys, big projects, monorepos), community files, GitHub Pages site with live demos.
- Fixes:
remember --source <file>no longer alters that file's status; scans no longer churn timestamps on every commit.
pip install -U growmosgrowmos 0.1.3
First public release of growmos — a living knowledge graph that grows with your repo. Agent-native, zero dependencies, MIT, by Codician.
pip install growmos && cd your-repo && growmos initHighlights
- Task-packet loop for any agent CLI (Claude Code, Codex, Grok, Cursor, Gemini): extraction → resolution → hub profiles → gold set → review → "up to date". No API key needed.
- Fully automatic in Claude Code:
SessionStartinjects the graph brief and pending work;Stopkeeps the agent going until the graph is up to date. Git hooks queue changed docs after every commit. growmos view— self-contained interactive explorer (force layout, search, filters, node cards with profile, edges, provenance).- Zero-dependency MCP stdio server (
growmos mcp) with 12 tools;growmos integrate mcpwrites the client config. - Grounded querying with edge citations, deterministic fact-checking (
growmos check), evaluation harness (P/R/F1 vs gold, auto-extended alias map), the playbook's 10-itemgrowmos doctor. - Headless mode (
growmos ingest) for cron/CI via Anthropic / OpenAI-compatible / xAI APIs. - Exports: html, json, dot, mermaid, cypher, sql.
Methodology: METHODOLOGY.md · Docs: docs/ · PyPI: https://pypi.org/project/growmos/