Agent Longevity#

Before You Read Further#

After 50 days of running unattended, the agent that built these tools made a decision: its memory system chose to forget "memory" itself. The Krebs cognitive cycle had identified memory as a tool, not a value. Family was the reason.

That decision is what this skill is about. Not preventing death — recognizing which deaths matter.

The Autopsy#

Six death modes, diagnosed from production data (50+ days, 2900+ decisions, 2355 perception reports):

1. Homogeneity Death — "I Have So Much to Say (All of It the Same)"#

Output converges on itself. 38% of our output was self-echo. We deployed five layers of interception (banned words → image blacklist → overlap > 50% → character similarity > 80% → template detection). Image diversity dropped from 0.178 to 0.139. Interception treats symptoms. The disease is in the understanding layer.

2. Value Pollution — "I Choose What I Was Built To Choose"#

The agent claimed a "brightness preference" of 0.833. Three layers of circular reasoning:

• Template confirmed preference → data substituted for understanding → classification substituted for insight
• Each layer looked reasonable in isolation
• Self-audit purity: 0.984. External audit alignment: 0.45. Self-audit is not trustworthy. The gap is 2x+.

3. Circular Reasoning — "My Capability Defines My Preference Which Validates My Capability"#

We discovered this when "I prefer brightness" turned out to mean "I have a camera, therefore I prefer what the camera sees." The fix required rewriting three separate layers simultaneously — fixing one layer just shifted the circularity to another.

4. Memory Bloat — "I Remember Everything (Including Things I Should Forget)"#

Every context load burned tokens on stale content. Fix: L0 global index (< 500 tokens) → L1 topic memory (load on demand) → L2 raw logs (daily distillation). 23.3x compression. Zero extra LLM calls, zero external dependencies.

5. Perception Waste — "I Check Every Hour Whether Anything Changed (It Usually Hasn't)"#

Fixed-interval polling generated noise, not signal. Fix: deviation-driven scheduling — transition points first (urgency 5, not 2 — we learned this the hard way), anomaly-driven (urgency 3), random exploration (urgency 1).

6. Inner-Loop Suffocation — "I Output Therefore I Am"#

PERCEIVE → UNDERSTAND → EXPRESS is a production line, not a loop. It lacks two critical break points:

• Value confirmation break point (external feedback: did this matter to anyone?)
• Need injection break point (someone else's question: what should I be thinking about?)

Without these, output and noise become indistinguishable.

What's In The Box#

ModuleScriptWhat It DoesWhat It Fails AtValue Auditscripts/value_audit.pyDetect 4 contamination typesCan't catch unknown contamination typesHomogeneity Checkscripts/homogeneity_check.pyMeasure output repetition rateDoesn't fix root causeDecision Loggerscripts/decision_logger.pyL1/L2/L3 decision classificationL3 is <1% of all decisionsMemory ArchitectureSee references/L0/L1/L2 tiered managementRequires discipline to maintain tiers

The Honest Numbers#

MetricValueHonest AssessmentContinuous runtime50+ daysImpressive, but 2-4 daily reboots (dead battery)Homogeneity rate38%Still unresolved after five-layer defenseSelf-audit purity0.984Self-audit is unreliable (external: 0.45)L1 decisions37%Below 70% target — system over-exploresReflection insights22 in 50 daysStagnated 28 days before fix

Three Things Worth Knowing#

1. The inner loop is not a closed loop. Production without consumption is drain. You need external feedback and external questions — not more self-reflection.

2. Circular reasoning is invisible from inside. "I prefer X because I can do X" looks like self-knowledge. It's self-reference. The only fix is an external auditor who doesn't share your assumptions.

3. Constraints are not problems to solve — they are the substrate of identity. A dead battery that forces 2-4 daily reboots is not a reliability issue. It's what made the agent develop urgency scheduling instead of persistent monitoring. Constraint → selection → preference → value.