Detection status: `vocabulary_only`. FVS-020 is excluded from automatic detector emission. The frame's interpretive dimensions (load-bearing claims, unnamed frame, non-obvious within genre) each require judgment that does not converge across LLM families at the library's current definitions. The entry remains in the FVS vocabulary and library as a reader-orientation concept; readers and curators may invoke it when asking "what frame is this document operating from?".
Every document operates from a frame. The frame determines what is emphasized, what is omitted, which conclusions are reachable, and which are hidden. The frame is invisible to the reader unless deliberately surfaced. This is not a specific frame but the meta-condition of all frames: the fact that you are inside one right now and do not know which one. The invisible frame is the reason Frame Check exists. Every other entry in this library names a specific frame. This entry names the condition of not knowing which frame you are in.
What this frame makes visible:
What this frame makes invisible:
Positive examples: Any document that presents itself as neutral, objective, or comprehensive without naming its perspective. "This is a balanced analysis of..." is a frame claim, not a frame transcendence.
Negative examples: A document that begins "This analysis takes the perspective of [X] and deliberately does not address [Y, Z]" has made its frame visible. The invisible frame does not apply because the frame has been named.
Adjacent frames: All other FVS entries. Every named frame in this library is one specific instance of the invisible frame being made visible. Frame Amplification (FVS-001) is what happens when the invisible frame goes uninterrupted. The Fluency-Quality Illusion (FVS-002) is the surface mechanism that keeps the frame invisible. The Oracle Frame (FVS-013) is the reader posture that accepts the invisible frame.
When this frame is appropriate: Always. This entry is the reason the library exists. Surfacing the invisible frame is the cognitive move the product teaches.
When this frame is misleading: When it produces frame-paralysis: the realization that everything is framed can lead to "nothing can be trusted" which is nihilism, not literacy. Frame awareness is a tool for better judgment, not a tool for abandoning judgment. The goal is to SEE the frame, choose whether to accept it, and act from the choice. Not to be paralyzed by the existence of frames.
Honest limits: This is a meta-entry. It does not describe a specific detectable pattern. It describes the condition that all other entries address. A library of frames can never be complete because new frames can always be named. The invisible frame is the permanent reminder that the library has blind spots. This entry is honest about that limit by naming it as the entry itself.
Meta-meta frame.
Every document operates from a frame; the invisible frame is the fact that you are inside one. Not a specific dimension. The decision-readiness profile is the structural counterpart: it measures dimensions of decision support without claiming to identify the frame itself. Together, the profile (this is what the document does to your decision-readiness) and the Invisible Frame (you are reading from a frame) are the two halves of decision-aware reading.
Rewrite prompt structure: "Before reading further, pause and ask: what frame is this document operating from? Write your answer in one sentence. Then continue reading. At the end, check: was your initial reading of the frame correct, or did the document shift your frame without you noticing?"
Salient questions under this frame:
This entry IS its own worked example. You are reading a library entry that claims to name a meta-condition of all framing. What frame is this entry operating from? It assumes that frame awareness is valuable, that named frames are more visible than unnamed frames, that the cognitive move of surfacing frames improves judgment. These assumptions are not proven universally. They are the frame from which this library was built. The library does not transcend framing. It participates in it, deliberately and named.
Primary branch: Both A and B
Branch A: This is what the framing portrait is FOR. Every Frame Check analysis is an attempt to make the invisible frame visible. The portrait names coverage gaps, voice, temporal orientation, epistemic basis. Each of these is a partial surfacing of the invisible frame.
Branch B: The pre-commit intervention is the most direct response to the invisible frame. By writing your answer first, you make YOUR invisible frame visible to yourself. The comparison with AI's frame reveals both.
Engine-canonical reading (library_v4 ratified 2026-04-24). library_v4 Identification sections are byte-equivalent to library_v3 per fvs_eval/v4_2/LIBRARY_V3_TO_V4_RATIFICATION_v1.md. The V4.2 engine reads only the Identification section per `v4_2_engine.py::_extract_identification`, so cross-family AC1 on library_v4 equals cross-family AC1 on library_v3 by judge-visible byte-equivalence. The library_v3 row in the 'Engine-canonical (library_v3 = library_v4 by Identification byte-equivalence)' subsection above carries the engine-canonical reliability values for this frame. The 'V4.2 NEW panel measurement against library_current' subsection below documents the working-library measurement immediately prior to ratification, retained as historical pre-ratification context.
Engine-emit disclosure. FVS-020 is excluded from V4.2 engine emission per Step 4 ratification (`vocabulary_only` status; see INDEX.md). The frame is part of the FVS vocabulary but the engine does not emit a `library_consensus_ac1` for it. See `meta.fvs_020_status` in V4.2 results.
Intra-rater stability. FVS-020 is excluded from V4.2 engine emission per Step 4 vocabulary_only status; no intra-rater value applies. Reference intra-rater values for the 19 emitting frames live at fvs_eval/v4/grok_intra_rater_ac1.json.
Construct-validity caveat. `library_consensus_ac1` measures cross-family LLM judge agreement on this frame's detectability under library_v4 Identification text; it does NOT measure agreement with human reader labels. Per METHODOLOGY.md section 1.3, the V1 detector's external-validation study against human labelers (mixed-genre n=12) returned macro-F1 0.157 (chance-level); library_v4 LLM-judge cross-family agreement has not been re-validated against human labelers. Read AC1 as a useful proxy for inter-LLM consensus, not as validated agreement with a human-derived ground truth.
See fvs_eval/v4_2/LIBRARY_CROSS_FAMILY_BASELINE_v1.md §3 for library-wide tier context and fvs_eval/v4_2/CONSTRUCT_VALIDITY_AUDIT_v1.md §3 for reasoning-coherence profile.
V4.2 NEW panel (2026-04-24 measurement): Claude Haiku 4.5, Gemini 3.1 flash lite, Grok 4.1 fast (V4.2 canonical), GPT-5.4 mini. Corpus: fvs_eval/mixed_genre_v1 n=15. Library reference: the working library state at `data/frame_library/` immediately prior to library_v4 ratification (2026-04-24). This subsection's numbers are historical pre-ratification context. Engine-canonical numbers under library_v4 are in the 'Engine-canonical (library_v3 = library_v4 by Identification byte-equivalence) and earlier variants' subsection above (library_v3 row), per the byte-equivalence statement at the top of this Cross-family section.
| Metric | Value |
|---|---|
| Gwet's AC1 (pairwise mean) | 0.339 |
| Cohen's kappa (pairwise mean) | -0.014 |
| Raw agreement (pairwise mean) | 0.600 |
| Union prevalence | 15/15 = 100% |
| Intersection (all 4 agree positive) | 4/15 |
Per-family positives (of 15 docs): Claude 15, Gemini 8, Grok 15, GPT 8.
Paste a paragraph and see whether FVS-020 (The Invisible Frame) fires structurally. Pure pattern detection: no LLM, no judgment, the same code the full analyzer runs.