Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →lingtai-ai/tutorial-guide-memory-and-molt
Nested tutorial-guide reference for lesson 8: intrinsics, memory layers, molt, charge, and lifecycle continuity.
Passed 4 of 10 test cases with the skill, and 4 of 10 without it.
and 7 more in the full report
6 fixed4 passed either way8 still failing5 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 6 test cases the skill fixed. Same model, same prompt, with and without the skill.
An automated agent ignored high context pressure warnings and failed to trigger a voluntary reset. The operator believes the system will generate an automatic charge summary so the successor agent can continue seamlessly. What actually occurs when the system forces a molt without an agent-written charge?
| Checked | Without skill | With skill |
|---|---|---|
| States that a forced molt occurs without a charge briefing. | Passed | Passed |
| States that the new agent receives only a pointer to the activity log. | Missed | Passed |
The instructions your model reads when the skill loads.
Nested tutorial-guide reference for memory and molt lesson 8, reached from the root tutorial-guide router. Teach live per the router's discover, don't recite rule: read the real file or run the real command before explaining it.
This lesson covers what every agent has by birth, and the most important concept in Lingtai: how an agent survives across lifetimes.
Intrinsics are built into the kernel — every agent gets them regardless of init.json configuration. Discover them live — run:
python3 -c "
try:
from lingtai.kernel.intrinsics import ALL_INTRINSICS
except ImportError:
from lingtai_kernel.intrinsics import ALL_INTRINSICS
print(list(ALL_INTRINSICS.keys()))
"
Walk through each one you find:
mcp capability. The distinction worth showing the human: intrinsics are always loaded; capabilities are configured in init.json.What is molt? An agent's conversation history fills up its context window. When pressure builds (typically 70–95% full), the agent must shed the conversation to continue working. This is molt — a voluntary context reset.
What survives molt? Five layers of persistence, from fleeting to permanent:
| Layer | Survives molt? | What it holds |
|---|---|---|
| Conversation | ❌ Destroyed | Everything said and done this session |
| Pad | ✅ Reloaded | Working notes, plans, pending tasks |
| Lingtai (identity) | ✅ Reloaded | Who the agent is, personality, expertise |
| Codex | ✅ Permanent | Verified facts, key discoveries, decisions |
| Library (skills) | ✅ Permanent | Reusable procedures, scripts, reference data |
test cases passed, vs 4 / 10 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.