Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/xxyoudeadpunkxx/shaping-frame-for-claude/llms.txt

Use this file to discover all available pages before exploring further.

Shaping Frame reduces decision drift by creating friction at epistemic thresholds. It does not eliminate drift. Understanding its limits is part of using it correctly. The limits below are not edge cases discovered in practice — they are declared properties of the design.
“The frame reduces drift by creating friction at points where unvetted material is becoming a decision. It does not eliminate drift — it makes it visible.”
The skill does not remember previous sessions. Each conversation starts fresh. The weight-tracking system is reconstructed from the current session context every time — there is no database, file, or model-level memory carrying state across conversations.The only carry-over mechanism is OP-historical: OP statements, corrections, or selections that are embedded inside imported EXT material. OP-historical carries higher weight than generic EXT content, but it is not equivalent to a current OP decision. If it grounds a new or structural decision, the skill requires you to surface and reconfirm it before proceeding — because context shifts between sessions.
Shaping Frame can flag material as Needs verification, but it cannot verify facts itself. The actual check must come from a file, tool, web source, or explicit OP decision.When an element is flagged Needs verification and verification is not possible in the current context, the element is frozen — it cannot advance state or ground a decision. It is surfaced explicitly at the next surface trigger. The skill marks the boundary; it does not cross it on your behalf.
The skill is not a background process or interceptor. It operates turn by turn within a conversation, reconstructing its operating frame from active session context. There is no hook that fires independently of Claude’s conversation loop, and no persistent runtime guard outside the current session.This means the skill’s effectiveness depends on the session context remaining active and the verbatim-read requirement being respected at load time.
The pre-write gate fires before structural writes, but Shaping Frame does not manage files, directories, or project structure. It operates in the semantic layer — tracking the epistemic state of proposals and decisions, not the filesystem state of outputs.File management, code organization, and project structure are outside its scope. The gate asks whether the content is ready to persist; it does not control how or where the content is persisted.
The skill does not make decisions. It tracks weight and surfaces thresholds. Only OP crystallizes decisions.Claude tracks the movement of material through Spark → Candidate → Tracked → Crystallized or Rejected. Claude emits checkpoints when thresholds are crossed. But the checkpoint is always a question — never an answer. No element advances to Crystallized without an explicit OP action.
The cognitive checkpoint operates in the semantic layer. It depends on threshold clarity and consistent application across every turn of a conversation.In long or fast sessions — where scope shifts rapidly, many elements are in motion at once, or the conversation compresses multiple decisions into a short exchange — some threshold crossings will be missed. That is expected behavior, not a failure. The goal is to catch the structurally important crossings: the ones where an unvetted element would become an architectural constraint, a specification input, a naming decision, or a rule.
The frame reduces drift by creating friction at the most important thresholds. It does not catch every promotion of unvetted material. Silent accumulation — where a proposal is referenced twice, then three times, and starts behaving as if crystallized — can still occur, especially when the session context is compressed or when the model’s attention is directed elsewhere.The skill is designed to be lightweight and non-disruptive. It surfaces only when something crosses a threshold or when the conversation needs a snapshot. Attempting to catch every crossing would require interrupting flow constantly, which would make the skill unusable in practice.
The declared limits above are part of the design — not failure modes. Shaping Frame is intentionally lightweight and non-disruptive. Completeness would require interrupting conversational flow on every element state change, which would defeat the purpose of using it in working sessions. The skill is designed to catch the structurally important threshold crossings while staying quiet when nothing critical is at risk.

Build docs developers (and LLMs) love

Get started for freeTalk to us