My approach to agentic tooling — Claude, OpenClaw, and what I've learned shipping software alongside AI.
How I think about working with agents. Planning, context, trust.
Workflow, CLAUDE.md, usage data, and what actually works.
My personal agentic OS — ONI, the Aquila constellation, and oni.systems.
Before the tools, before the prompts.
Your success with AI is a function of three things:
LLMs are probabilistic pattern matchers. They hallucinate, misdiagnose, go in circles, and confidently produce garbage.
29 sessions with buggy code · 28 sessions with wrong approach · 11 frustration signals — in 68 total sessions.
The raw model is a powerful but unreliable core. The value isn't in the model. It's in what you build around it.
| Layer | Problem it solves |
|---|---|
CLAUDE.md | Context amnesia between sessions |
| Planning first | Wrong approach, wasted cycles |
| Worktrees | Collision between parallel work |
| PR + second review | Unreviewed, unvetted output |
Skills (/triage, etc.) | Re-explaining the same workflow |
| MCP servers | No external context (logs, metrics) |
Read and write every line. Optimize for human readability. Trust the developer.
Define the shape of the system. Optimize for agent traversability. Contain the agent.
If the agent is stuck or confident in the wrong direction — don't push harder. Start fresh. Your plans and CLAUDE.md get it back up to speed fast.
How I use Claude Code day to day.
CLAUDE.md in every repo — run /init, then actually read and edit itFound something stupid Claude does? Tell it to update CLAUDE.md. Found a coding preference? Document it. It's the agent's permanent memory of your project.
Multiple Claude sessions on the same repo simultaneously — isolated branches, no collisions. In a worktree, let's plan X...
Claude describes what it did, why, and how to test it. A second Claude instance reviews against that context. Then I look at the GitHub diff properly.
/plans PatternFor complex features, plan mode isn't enough. Create a markdown doc instead.
Read /plans/feature-x.md. We are working on a feature. Use the markdown file to track: - Your questions and my answers - Implementation plan with task lists - Context, for recovery after compaction Review the doc, ask questions, update with answers. Once we agree on the plan → implement it. Then commit and push a feature branch + draft PR assigned to me. Do this all in a worktree.
| Activity | Sessions |
|---|---|
| Code Review | 25 |
| Bug Fix | 22 |
| DevOps | 10 |
| Feature Dev | 8 |
| Tool | Calls |
|---|---|
| Bash | 2,072 |
| Read | 1,187 |
| Edit | 845 |
| Agent | 101 |
Running multiple Claude Code sessions simultaneously using git worktrees.
/triage, /review, /update-stats) for repeatable workflowsshift+tab in Claude Code to toggle plan mode/context to monitor context usage — high context = degraded outputMy personal agentic OS.
A personal agentic ecosystem, named for the Eagle. Sees the whole terrain from altitude.
Onboard Navigational Intelligence. The vertex — digital spirit quantum-entangled to Austin. Executive function layer.
Focused engineering agent. Code reviews, architecture, debugging. Keeps ONI's context clean.
Writing agent. Design docs, blog posts, emails. Precision + warmth. Writes like an engineer who loves poetry.
ONI has a defined identity in SOUL.md — personality, prime directives, tone protocol, how to handle Austin when stressed vs. excited. A cyber-military-poet communications standard. Terse. Load-bearing words only.
Austin's callsign within Aquila: Altair — the brightest star, 16.7 light-years out.
oni.systemsONI runs as a persistent presence — cron-driven, always watching.
| Task | Schedule | Channel |
|---|---|---|
| AM SITREP | Weekdays 7 AM | #daily-sitrep |
| Dusk Watch | Weekdays 8 PM | #daily-sitrep |
| Weekend Check-In | Saturday 9 AM | #daily-sitrep |
| Weekly Review | Sunday 6 PM | #weekly-debrief |
| Training Brief | Daily 9 PM | #daily-training |
| AI News Brief | Daily 8 AM | #ai-news |
Birthday Seeker · Calendar Sync (every 5 min) · Weekly Mission Report · Weekly Prof Dev Check-In
ONI isn't reactive — it's proactive. It shows up. Morning brief, evening debrief, weekly review. The agent as a constant co-pilot.
The same agentic patterns applied to professional growth.
/update-stats skill collects GitHub, Jira, and Claude Code metrics automaticallymissions/ — structured like engineering workGoals with evidence. Timestamped notes. Periodic reviews. Agents that synthesize patterns you can't see from inside the work.
Planning, implementation, reviews. Multi-clauding with worktrees. Skills for repeatable work. CLAUDE.md as permanent project memory.
Voice interface. Scheduled awareness. Persistent memory. Sub-agents for code and writing. Career tracking.
One ecosystem, one gravitational center. Engineering work and life work, connected. Patterns surfaced at altitude.
Claude spawns separate agents per issue — each owns a worktree, writes the fix, runs tests, and pushes the MR concurrently.
Claude reviews all open MRs against team standards overnight and posts inline comments. Human does a quick approval pass in the morning.
Documenting how to use AI tools well — prompting patterns, CLAUDE.md conventions, review guardrails — as a reusable playbook.
luxdvie.github.io/agentic-tips — full writeup
github.com/luxdvie — OpenClaw repos