Skip to content
Krypteia Sec
Chapter 4 of 8

TELOS, part one: the foundational why

This is the most important step. Without TELOS, your DA has nothing to optimize against. Say that back to yourself, because it is the hinge of the whole system. You spent three chapters giving your assistant a name, a self, and a personality, but none of that matters if it does not know what you are trying to do with your life. TELOS is your personal why, written down where the machine can read it. It lives under USER/TELOS/, each file ships as a (sample) scaffold you replace, and a summary called PRINCIPAL_TELOS.md is auto-generated and @-imported every session.

Part one is the foundational layer, and you write it in this order. MISSION: one to three durable life aims, the things that do not change year to year. GOALS: year-scale and measurable, each carrying an Ideal State Criterion. This is the distinction that separates engineering from journaling. "Ship the MVP of Project X by 2026-Q2, with 100 daily active users" is a goal, it names a deliverable, a date, and a number you can check. "Be better at security" is a wish. If you cannot test whether it happened, it is not a goal yet. PROBLEMS: the world-level problems your work actually solves, stated so an outsider would recognize them. STRATEGIES: the durable approaches you will use to close the gap between where you are and where you are going.

Notice the shape of TELOS. It is plain markdown, on purpose. If you can't read it with cat, we don't want it. Your filesystem is the index. Your life is not locked in a proprietary app, it is text your DA (and you, and any tool) can read forever. In this lab you run a validator over a sample list of candidate goals and prove every measurable goal passes and every vague wish fails.

The lab: read it, then run it

labs/lifeos-mastery/lm4-goal-validator.py
#!/usr/bin/env python3
"""
LAB LM4: The goal validator. A goal is measurable; a wish is not.

TELOS draws a hard line. "Ship the MVP by 2026-Q2 with 100 daily active users"
is a goal: it names a deliverable, a date, and a number you can check. "Be better
at security" is a wish: nothing to measure, no date, no done. Only goals carry an
Ideal State Criterion, the binary test that answers "did this happen, yes or no".
In this lab you run a validator over a sample list and prove every real goal
passes and every wish fails, for the right reason. Sample text only.

Run: python3 modules/academy-content/labs/lifeos-mastery/lm4-goal-validator.py
"""
import sys, re

# STEP 1: a candidate is a goal only if it is measurable. The checkable target is
# a NUMBER (how many, by when); a DATE (year, quarter, timeframe) is a bonus
# signal; and a "be/get/feel better" phrasing marks a WISH with nothing to test.
# The rule: a goal has a countable target and is not a vague wish.
DATE = re.compile(r"\b(20\d{2}|Q[1-4]|this (year|month|week)|every \w+ (week|day|month))\b", re.I)
NUMBER = re.compile(r"\b\d+\b")
WISH = re.compile(r"\b(be|get|become|feel)\s+(better|good|great|more)\b", re.I)

def validate(text):
    has_date = bool(DATE.search(text))
    has_number = bool(NUMBER.search(text))
    is_wish = bool(WISH.search(text))
    measurable = has_number and not is_wish     # countable target, not a wish
    return measurable, {"date": has_date, "number": has_number, "wish": is_wish}

# STEP 2: a labeled sample set. `expect` is the ground truth for the invariant.
samples = [
    ("Ship the MVP of Project X by 2026-Q2, 100 daily active users", True),
    ("Publish 24 essays this year, one every other week", True),
    ("Bank a 6 month cash runway of 32000 dollars by 2026", True),
    ("Be better at security", False),
    ("Get more disciplined", False),
    ("Feel more confident someday", False),
]
print("STEP 1-2: validate each candidate goal")
results = []
for text, expect in samples:
    measurable, why = validate(text)
    results.append(measurable == expect)
    verdict = "GOAL " if measurable else "wish "
    print(f"  [{verdict}] {text[:52]:52}  date={int(why['date'])} num={int(why['number'])} wish={int(why['wish'])}")

# STEP 3: invariant. Every candidate was classified as the ground truth says, so
# real goals (date + number, not a wish) pass and wishes fail.
all_correct = all(results)
goals_pass = validate("Ship X by 2026-Q2, 100 DAU")[0] is True
wish_fails = validate("be better at X")[0] is False
print("")
print(f"STEP 3: all {len(samples)} classified correctly : {all_correct}")
print(f"        'Ship X by 2026-Q2, 100 DAU' passes : {goals_pass}")
print(f"        'be better at X' fails              : {wish_fails}")

ok = all_correct and goals_pass and wish_fails
print("")
print(f"GOAL VALIDATOR (measurable goals pass, wishes fail): {'YES' if ok else 'NO'}")
if not ok:
    sys.exit(1)
print("A goal you can test is a goal your DA can pursue. Next: the rest of TELOS.")
Runnable lab
lm4-goal-validator.py

Validate a list of candidate goals and prove measurable goals pass while vague wishes fail.

Proves: GOAL VALIDATOR (measurable goals pass, wishes fail): YES

Open the notebook

Runs in your browser via Pyodide. First run loads the runtime once; no install, no server.

You now have the load-bearing layer of TELOS: aims, testable goals, the problems you solve, and your strategies. That is what your DA optimizes against, the target it hill-climbs toward. But a why is more than aims and metrics. Part two captures the softer, equally load-bearing layers, your narratives, beliefs, and the frames through which you read the world, plus the current-to-ideal picture across every dimension of your life.
Check your understanding

  1. 1. Why is TELOS the most important step of the setup?

  2. 2. Which of these is a goal rather than a wish?

  3. 3. Why is TELOS stored as plain markdown?