agentleFS
madeinoz67madeinoz67/madeinoz-knowledge-system/CLAUDE.md

madeinoz-knowledge-system

CLAUDE.md3 starsChanged 10 months agoMITView on GitHub

Personal knowledge management system using Graphiti knowledge graph with Neo4j/FalkorDB. LLM Compatibility: - Working: gpt-4o-mini, gpt-4o, Claude 3.5 Haiku, Gemini 2.0 Flash - Failing: All Llama/Mistral variants (Pydantic validation errors) After modifying docker/patches/*.py, you MUST rebuild: If changes don't appear: Docker caching issue. Stop containers fully before restart. MUST use --profile development when testing against dev containers. Rules: Never reconfigure profiles. Always pass --profile development for dev work. ⚠️ NEVER use gh release create directly - it creates the tag…

  • Commits and pushes

The file

We split this CLAUDE.md into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    After modifying docker/patches/*.py, you MUST rebuild:

    line 57
  2. Very specific:

    Rules: Never reconfigure profiles.

    line 85

≈ “Common” is an estimate: at least 1 in 5 of the 27,641 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

madeinoz-knowledge-system Development Guidelines

Personal knowledge management system using Graphiti knowledge graph with Neo4j/FalkorDB.

Quick Reference

ResourceLocation
Full docsdocs/ or https://madeinoz67.github.io/madeinoz-knowledge-system/
Memory Decaydocs/usage/memory-decay.md
Configurationdocs/reference/configuration.md
CLI Referencedocs/reference/cli.md
Observabilitydocs/reference/observability.md

Commands

bun run build              # Build TypeScript
bun run typecheck          # Type check only
bun test                   # Run tests
bun run server-cli start   # Start containers
bun run server-cli stop    # Stop containers
bun run server-cli status  # Check status

Architecture

src/
├── server/           # MCP server orchestration, container management
├── skills/           # PAI Skill definition and workflows
├── hooks/            # Session lifecycle hooks (memory sync)
└── config/           # Environment configuration

docker/
├── patches/          # Python MCP server code (graphiti patches)
└── Dockerfile        # Container build

config/
├── decay-config.yaml # Memory decay settings (180-day half-life)
└── sync-sources.json # Memory sync configuration

Technical Details

  • Runtime: Bun (ES modules, target: bun)
  • TypeScript: Strict mode, path aliases (@server/*, @lib/*)
  • Database: Neo4j (default, port 7474/7687) or FalkorDB (port 3000/6379)
  • Python: 3.11+ in container for MCP server patches

LLM Compatibility:

  • Working: gpt-4o-mini, gpt-4o, Claude 3.5 Haiku, Gemini 2.0 Flash
  • Failing: All Llama/Mistral variants (Pydantic validation errors)

Container Development (CRITICAL)

After modifying docker/patches/*.py, you MUST rebuild:

docker build -f docker/Dockerfile -t madeinoz-knowledge-system:local .
bun run server-cli stop
bun run server-cli start --dev

If changes don't appear: Docker caching issue. Stop containers fully before restart.

CLI Profile Usage (CRITICAL)

MUST use --profile development when testing against dev containers.

# CORRECT - testing against dev containers
bun run ~/.claude/skills/Knowledge/tools/knowledge-cli.ts --profile development health
bun run ~/.claude/skills/Knowledge/tools/knowledge-cli.ts --profile development run_maintenance

# WRONG - omitting profile hits PRODUCTION data
bun run ~/.claude/skills/Knowledge/tools/knowledge-cli.ts health
ProfileContainerUse Case
production (default)ProductionLive data
development--dev containersTesting

Rules: Never reconfigure profiles. Always pass --profile development for dev work.

Releases (CRITICAL)

⚠️ NEVER use gh release create directly - it creates the tag via GitHub API without triggering the CI workflow. The release job will never run, leaving you with a broken release (no Docker image pushed, no docs deployed).

Release checklist - follow these steps exactly:

# === PRE-RELEASE CHECKS ===
# 1. Check you're on main branch
git branch --show-current  # Should output: main

# 2. Pull latest changes
git pull origin main

# 3. Verify working tree is clean (no uncommitted changes)
git status  # Should show: "nothing to commit, working tree clean"

# 4. Verify latest CI passed
gh run list --limit 1  # Check status is "completed success"

# === CREATE RELEASE ===
# 5. Create annotated tag with release description
git tag -a v1.x.x -m "Release v1.x.x" -m "Description of what's in this release"
# Or for multi-line descriptions, omit -m to open your editor:
git tag -a v1.x.x
# Then write your description in the editor (same format as commit messages)

# 6. Push tag to trigger CI workflow
git push origin v1.x.x

# === VERIFY RELEASE ===
# 7. Watch workflow run
gh run watch

# 8. Confirm release created
gh release view v1.x.x

What the CI workflow does automatically:

  • Generates changelog from git-cliff using conventional commits
  • Updates CHANGELOG.md on main branch
  • Builds and pushes multi-arch Docker image to GHCR
  • Creates GitHub Release with formatted release notes
  • Deploys documentation to GitHub Pages

Recovering from a bad release:

# If you accidentally used gh release create or created via web UI:
gh release delete v1.x.x --yes
git tag -d v1.x.x
git push origin :refs/tags/v1.x.x
# Then follow the correct process above

Why this matters: The release job (.github/workflows/ci.yml:195) only triggers on push events for tags matching v*. Tags created via GitHub web UI or gh release create generate a ReleaseEvent, not a push event, so the workflow never runs.

Active Technologies

  • JSON (Grafana dashboard configuration) + Grafana (provisioned), Prometheus (data source), existing metrics from PR #34 (013-prompt-cache-dashboard)
  • N/A (dashboard configuration only) (013-prompt-cache-dashboard)
  • Python 3.11+ + OpenTelemetry Prometheus exporter, Graphiti MCP server (015-memory-access-metrics)
  • Neo4j / FalkorDB (existing graph database) (015-memory-access-metrics)
  • PromQL (Prometheus Query Language), JSON (Grafana dashboard format) + Grafana 10.x, Prometheus (scraping OpenTelemetry metrics) (016-prometheus-dashboard-fixes)
  • Python 3.11+ + OpenTelemetry Prometheus exporter, threading.Lock for thread safety, Grafana dashboard JSON (017-queue-metrics)
  • Python 3.11+ (MCP server), TypeScript/Bun (CLI) + FastMCP (Python), Graphiti Core (knowledge graph), mcp-client (TypeScript) (020-investigative-search)
  • Neo4j (default) or FalkorDB with graph traversal suppor (020-investigative-search)

Recent Changes

  • 017-queue-metrics: Queue processing metrics for monitoring queue depth, latency, consumer health, and failure tracking
  • 013-prompt-cache-dashboard: Added JSON (Grafana dashboard configuration) + Grafana (provisioned), Prometheus (data source), existing metrics from PR #34

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.