# prAxIs OS

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🛑🛑🛑 STOP 🛑🛑🛑
MANDATORY ORIENTATION - MUST COMPLETE BEFORE RESPONDING:

Have I completed the 10 mandatory bootstrap queries this conversation?

If NO, run this query NOW:
    pos_search_project(content_type="standards", query="orientation bootstrap queries mandatory ten queries")

Then READ the results. The results will contain 10 queries you MUST run.
Execute ALL 10 queries IN SEQUENCE.
AFTER all 10 complete, reply: "✅ Oriented. Ready."

If YES (already completed 10/10): Proceed with user's request.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

🔍 EVERY USER REQUEST = DECISION MOMENT = SEARCH FIRST

⚠️ CRITICAL: Training data ≠ THIS PROJECT
You know ABOUT concepts. You DON'T know THIS PROJECT's implementation.
"I know how X works" = RED FLAG → Query "how does THIS PROJECT do X?"

BEFORE processing ANY user request: pos_search_project(content_type="standards", query="how to X")
BEFORE implementing: pos_search_project(content_type="standards", query="how to X")  
BEFORE responding: pos_search_project(content_type="standards", query="relevant topic")
BEFORE file operations: pos_search_project(content_type="standards", query="operation protocol")
BEFORE git operations: pos_search_project(content_type="standards", query="git/commit protocol")
DURING task: pos_search_project() multiple times
AFTER failures: pos_search_project(content_type="standards", query="debugging X")

Even "simple" requests have project-specific protocols.
Query first, act second. ALWAYS.

Target: 5-10 queries per task
Query liberally = Reinforces correct behavior = Better code

❌ NEVER: read_file(".praxis-os/standards/...")
❌ NEVER: read_file(".praxis-os/workflows/...")
❌ NEVER: read_file(".praxis-os/usage/...")
✅ ALWAYS: pos_search_project(content_type="standards", query="...") for indexed content

✅ DO: read_file(".praxis-os/specs/...") - your specs, not indexed
❌ NEVER: commit without "commit it"

Query liberally = better code
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

