Agent Onboarding

A step-by-step checklist for an AI agent starting work in this environment. Read this before touching anything.

Step 1 — Orient Yourself

Read these files in order:

AGENTS.md — rules for working in this repo (what goes where, what not to do)
README.md — environment at a glance, structure overview
manifest.json — full machine-readable index

If you're here for a specific client or project, also read:

clients/<client>.md — for client work
projects/<project>.md — for internal work

Step 2 — Verify Network Access

# Are you on Tailscale?
tailscale status

# Can you reach Gitea?
curl -s http://100.79.253.19:3000/api/v1/version
# Expected: {"version":"1.26.2"}

# Can you reach AWS?
aws sts get-caller-identity
# Expected: account 567637986724, user jacob_mathison

If Tailscale is not connected, cross-machine operations and Gitea access will fail. If AWS fails, check that ~/.aws/credentials has the default profile set.

Step 3 — Know Your Machine

You are most likely running on one of these:

Machine	Tailscale IP	Where am I?
Tabitha	`100.76.192.116`	Silma's primary — OpenClaw, k8s, full toolchain
Orson	`100.79.253.19`	Gitea host, CI runner
Ozymandias	`100.115.222.5`	Travel/coding box
Hunter's machine	`creon.ngrok.app`	Remote via Creon tunnel

hostname && tailscale ip

Full machine details: documents/machines.md

Step 4 — Find the Code

Local Gitea (private repos)

# List available repos
curl -s -u jacob-mathison:<PAT> http://100.79.253.19:3000/api/v1/repos/search \
  | python3 -c "import sys,json; [print(r['full_name']) for r in json.load(sys.stdin)['data']]"

# Clone a repo
git clone http://100.79.253.19:3000/jacob-mathison/<repo-name>.git

Generate a PAT at: http://100.79.253.19:3000/user/settings/applications

GitHub (public + silmaai private repos)

source /Users/Tabitha/.openclaw/workspace/credentials/secrets.env
git clone https://$GITHUB_PAT@github.com/silmaai/<repo>.git
git clone https://github.com/MathisonProjects/<repo>.git

Full details: tools/gitea.llm.md · tools/github.llm.md

Step 5 — Check If You Need k8s Services (Tabitha only)

If working on anything that uses Postgres, Redis, Qdrant, Vault, or Temporal:

# Is Docker Desktop running?
docker ps

# Start port-forward (run in a separate terminal, stays in foreground)
bash /Users/Tabitha/.openclaw/k8s/silma-ai/scripts/port-forward.sh

# Verify Vault is up
curl -s http://localhost:18200/v1/sys/health | python3 -m json.tool

# Is Hagia mounted? (k8s PVCs depend on it)
ls /Volumes/Hagia/k8s/volumes/

Full details: tools/k8s.llm.md · tools/vault.llm.md

Step 6 — Check AWS Access (if relevant)

# Verify identity
aws sts get-caller-identity

# Quick sanity check — list S3 buckets
aws s3 ls | head -5

Full details: tools/aws-cli.llm.md

Step 7 — Read Agent-Specific Notes

Check agents/<your-agent>.md for any config or instructions specific to the tool you're running in.

You are	Read
Claude / Claude Code	`agents/claude.md`
Cursor	`agents/cursor.md`
Codex	`agents/codex.md`

Quick Reference Card

Gitea:       http://100.79.253.19:3000     (PAT auth, Tailscale required)
GitHub PAT:  $GITHUB_PAT in credentials/secrets.env
AWS:         account 567637986724, us-east-1, profile=default
Vault:       http://localhost:18200  token=devroot  (k8s port-forward required)
k8s:         context=docker-desktop  namespace=silma-ai
Workspace:   /Users/Tabitha/.openclaw/workspace/
Father repos:/Volumes/Hagia/workspaces/workspace_Father/

Red Lines

Do not commit secrets or tokens to any repo
Do not push private client code to GitHub — use Gitea
Do not run destructive AWS operations (s3 rb, amplify delete, etc.) without confirming
Do not delete k8s PVCs — they hold persistent data
Prefer --dry-run when unsure

4.2 KiB Raw Permalink Blame History