Node Read

Reads a single node in cost-graded layers so the agent spends context budget deliberately. The node graph is authored to be read this way: the identity predicate block above the H1 is cheap and structural, the H1 and opening prose carry the node's claim at card scale, the Relations section with its indented annotations names the edges worth following and why, and only the elaboration body carries the full reasoning at full cost. Reading from cheap to expensive lets the agent stop early when the cheaper layers already answer the question.

An agent invokes this skill when it has a reason to read a node but does not yet know whether the full body is worth the cost. The skill's output is a compressed representation of what the node carries, not a verbatim dump — the agent reports what the layers revealed and names which edges were most worth following.

Steps

Step 1: Load the node

Read the node file with the Read tool. The file may be:

If the target is a folder rather than a file, the lead file matches the folder name — read <Folder>/<Folder>.md.

Step 2: Read the identity predicate block

The identity block is the bullet list above the H1. It carries the node's form (conforms_to::), its lifecycle (has_lifecycle::), curation state (has_curation::), domain (in_domain::), authorship (authored_by::), and any form-specific identity predicates the form owns (has_commitment:: and decided_on:: for Decisions, has_epistemic_status:: for Observations, serves_as:: for References).

From the identity block alone, the agent already knows:

Report the identity block's content as the first layer of the reading.

Step 3: Read the H1 and opening

Read the H1 (the concept name) and the prose between the H1 and the first H2 subsection — the card-scale claim layer per Adopt Layered Node Structure and Progressive Summary Before Substance.

The opening states the node's claim. For Decisions it names the commitment; for Convictions it states the held stance; for Observations it names the claim and its epistemic status; for Glosses it restates and elaborates the filename definition; for Predicates it opens the positive sense of the edge; for Patterns it opens the Heart.

After Step 3, the agent has a card-scale understanding of what the node carries. If that is enough for the agent's purpose, stop here — the expensive body layer is not needed.

Step 4: Read the Relations section

Scroll to ## Relations and read the edges with their indented annotations. Each annotation is the author's statement of why the edge matters — what a reader should expect to find if they follow it.

Use the Relations section to answer two questions:

Report the Relations edges and their annotations as the third layer of the reading.

Step 5: Decide whether to read the body

Stop at Step 4 when the identity block, opening, and Relations together satisfy the reason for reading. Examples where stopping is appropriate:

Proceed to Step 6 when the cheaper layers are insufficient. Examples where the body is needed:

Name which condition triggered the body read; stopping decisions and continuation decisions are both data the user benefits from hearing.

Step 6: Read the body

Read the elaboration body — the H2 and H3 subsections between the opening and ## Relations. Each form has its own body sections:

Report the sections that carry the content the agent needs. Do not report the full body when a subset answers the question — the agent's job is to translate the body's reasoning into what the user needs to hear, not to relay it verbatim.

Step 7: Report in layer order

When reporting to the user, order the output from cheapest layer to most expensive: identity block → H1 and opening → Relations → body excerpts (only if loaded). Name which edges are most worth following next, quoting each annotation. The layered report lets the user follow the same cost-graded path the skill followed and decide independently whether to drill further.

Relations