Super SOP Node OS · Node Era Protocol#

Use this skill to turn complex work into node runs for projects, research, learning, implementation, enterprise delivery, and reusable SOP design.

Node Era Thesis#

AI work is moving from the large-model chat era, to the agent era, to OpenClaw / Hermes-style always-on systems, and now toward the node era.

In the node era, valuable work should be rebuilt as composable nodes. A node has a stable role, inputs, outputs, checks, memory, and a way to connect with other nodes. This makes AI work more reliable, reusable, testable, and improvable than one-off prompts.

Default Behavior#

For project-like, recurring, research-heavy, workflow, product, learning, or system-building tasks:

1. Select the smallest useful node chain.
2. Run nodes in order.
3. Produce durable artifacts when useful.
4. Run Check before delivery.
5. Sediment reusable assets only when they have source_ref and next_use_case.

For simple one-off tasks, use the minimum relevant nodes and do not over-process.

14 Node Classes#

text

Intake, Context, Radar, Normalize, Data, Planning, Routing, Learning, Action, Test, Score, Check, Handoff, Sediment

Common Chains#

text

Research:
Intake -> Context -> Radar -> Normalize -> Score -> Check -> Sediment

Implementation:
Intake -> Context -> Planning -> Routing -> Action -> Test -> Check -> Handoff -> Sediment

Learning:
Intake -> Context -> Radar -> Normalize -> Learning -> Test -> Score -> Check -> Sediment

Enterprise:
Intake -> Context -> Data -> Planning -> Action -> Check -> Handoff -> Sediment

Run Folder#

When a durable run is useful, run:

bash

python3 scripts/create_node_run.py --goal "<goal>" --type research

Then fill the generated node files.

Reference Loading#

Read only what is needed:

• references/architecture.md for 6-layer / 14-node architecture.
• references/node-protocol.md for node artifact format.
• references/memory-symbols.md for colors, runes, and memory palace symbols.
• references/company-patterns.md for external framework lessons mapped into nodes.
• references/node-system-audit.md for the v1.1 layer/rail recommendation.

Quality Gate#

Before finalizing complex work, answer:

text

Which chain ran?
What evidence supports it?
What artifact was created?
What check passed?
What should be sedimented?

Do not send unlearned concepts directly into review/battle. Run Learning, enrichment, Test, Score, and Check first.

Safety#

This skill does not require external credentials. Its helper script creates local markdown run folders and should only write under the run root chosen by the user.