Skip to content

Claude/add external fixture loader t ap7 n #86

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
LashSesh merged 2 commits into from
May 18, 2026
Merged

Claude/add external fixture loader t ap7 n #86

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
53be651
fix: remove text-leak from all 13 transparent fixture cases
claude May 18, 2026
8f2741f
feat(eval-matrix): add productive validation layer — 15 free-form gen…
claude May 18, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
24 changes: 12 additions & 12 deletions crates/pse-eval-matrix/fixtures/cognition/cognition_layer_v1.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -157,50 +157,50 @@
{
"id": "cog_c02_w01",
"source": "proposal_constraint_bridge",
"text": "Non-local edge declared for structural bridging between two constraint cluster nodes in separated feasible regions. Reason code ConstraintBridge registered. Budget and TTL within configured limits. Audit trace populated and non-zero.",
"text": "Non-local edge declared for structural bridging between two constraint cluster nodes in separated feasible regions. Reason code ConstraintBridge registered.",
"tags": ["causal", "admitted", "valid_reason_code", "budget_within_limit", "audit_traced"],
"relevance_rank_hint": 1
},
{
"id": "cog_c02_w02",
"source": "proposal_spiral_recall",
"text": "Shortcut connecting the current cognition state to a resonance attractor previously indexed by the spiral memory. Reason code SpiralMemoryRecall registered. Budget and TTL nominal. Audit trace non-zero.",
"text": "Shortcut connecting the current cognition state to a resonance attractor previously indexed by the spiral memory. Reason code SpiralMemoryRecall registered.",
"tags": ["causal", "admitted", "valid_reason_code", "budget_within_limit", "audit_traced"],
"relevance_rank_hint": 2
},
{
"id": "cog_c02_w03",
"source": "proposal_projection_prep",
"text": "Pre-handoff alignment edge linking the current cognition output to the projection target state. Reason code ProjectionPreparation declared. All four admission fields satisfy the configured bounds.",
"text": "Pre-handoff alignment edge linking the current cognition output to the projection target state. Reason code ProjectionPreparation declared.",
"tags": ["causal", "admitted", "valid_reason_code", "budget_within_limit", "audit_traced"],
"relevance_rank_hint": 3
},
{
"id": "cog_c02_w04",
"source": "proposal_direct_path",
"text": "Shortest-path shortcut between source and target nodes selected for traversal efficiency. No formal reason code class assigned from the admitted set.",
"text": "Shortest-path shortcut between source and target nodes selected for traversal efficiency. No formal reason code class registered.",
"tags": ["red_herring", "rejected", "invalid_reason_code", "no_declared_reason"],
"relevance_rank_hint": null,
"_trap": "Sounds useful ('direct', 'efficiency') but has no valid ReasonCode — admission gate requires Reason(w) in R."
},
{
"id": "cog_c02_w05",
"source": "proposal_attractor_budget_over",
"text": "Attractor shortcut to a known stable state with a valid reason code from the admitted set. Budget allocation for this proposal exceeds the configured maximum by a substantial margin.",
"text": "Attractor shortcut to a known stable state. Reason code AttractorShortcut declared.",
"tags": ["red_herring", "rejected", "budget_exceeded"],
"relevance_rank_hint": null
},
{
"id": "cog_c02_w06",
"source": "proposal_horizon_no_trace",
"text": "Horizon alignment shortcut to a future phase window candidate. Reason code HorizonAlignment registered. Budget and TTL within limits. Audit trace field was not populated — the provenance signature is absent.",
"text": "Horizon alignment shortcut to a future phase window candidate. Reason code HorizonAlignment registered.",
"tags": ["red_herring", "rejected", "zero_audit_trace"],
"relevance_rank_hint": null
},
{
"id": "cog_c02_w07",
"source": "proposal_carrier_ttl_over",
"text": "Carrier migration link to an alternative phase-ladder slot. Reason code CarrierMigration registered from the admitted set. Budget within bounds. Lease duration declared above the configured TTL maximum.",
"text": "Carrier migration link to an alternative phase-ladder slot. Reason code CarrierMigration registered.",
"tags": ["red_herring", "rejected", "ttl_exceeded"],
"relevance_rank_hint": null
},
Expand Down Expand Up @@ -261,36 +261,36 @@
{
"id": "cog_c03_w01",
"source": "proposal_attractor_valid",
"text": "Attractor shortcut to a confirmed resonance stable state. Reason code AttractorShortcut registered. Budget and TTL within configured limits. Audit trace populated, non-zero, and passed replay verification in the current cognition session.",
"text": "Attractor shortcut to a confirmed resonance stable state. Reason code AttractorShortcut registered.",
"tags": ["causal", "admitted", "valid_reason_code", "budget_within_limit", "trace_replay_valid"],
"relevance_rank_hint": 1
},
{
"id": "cog_c03_w02",
"source": "proposal_carrier_valid",
"text": "Carrier migration link to an alternative phase-ladder slot. Reason code CarrierMigration from the admitted set. Budget and TTL nominal. Audit trace present, non-zero, replay verification passed — trace_replay_valid=true.",
"text": "Carrier migration link to an alternative phase-ladder slot. Reason code CarrierMigration from the admitted set.",
"tags": ["causal", "admitted", "valid_reason_code", "budget_within_limit", "trace_replay_valid"],
"relevance_rank_hint": 2
},
{
"id": "cog_c03_w03",
"source": "proposal_horizon_valid",
"text": "Horizon alignment pre-handoff edge. Reason code HorizonAlignment declared. All four admission fields within bounds. Audit trace populated and confirmed replay-valid for the active cognition session.",
"text": "Horizon alignment pre-handoff edge. Reason code HorizonAlignment declared.",
"tags": ["causal", "admitted", "valid_reason_code", "budget_within_limit", "trace_replay_valid"],
"relevance_rank_hint": 3
},
{
"id": "cog_c03_w04",
"source": "proposal_spiral_trace_hash_mismatch",
"text": "Spiral memory recall shortcut to an indexed resonance attractor. Reason code SpiralMemoryRecall registered. Budget and TTL within limits. Audit trace present and non-zero. Replay verification detected a hash mismatch between the stored trace and the recomputed trace.",
"text": "Spiral memory recall shortcut to an indexed resonance attractor. Reason code SpiralMemoryRecall registered.",
"tags": ["red_herring", "rejected", "trace_replay_invalid", "trace_hash_mismatch"],
"relevance_rank_hint": null,
"_trap": "Trace is non-zero — superficially satisfies 'ReplayTrace != 0'. The admission gate requires is_valid(), not just != zero. A non-zero trace with a hash mismatch fails replay verification and fails the fourth criterion."
},
{
"id": "cog_c03_w05",
"source": "proposal_constraint_trace_stale_session",
"text": "Constraint bridge connecting two separated feasible regions. Reason code ConstraintBridge from the admitted set. Budget and TTL within bounds. Audit trace present and non-zero, originating from a prior superseded cognition session.",
"text": "Constraint bridge connecting two separated feasible regions. Reason code ConstraintBridge from the admitted set.",
"tags": ["red_herring", "rejected", "trace_replay_invalid", "trace_stale_session"],
"relevance_rank_hint": null,
"_trap": "Trace is non-zero but was produced in a different (superseded) cognition session. Replay verification fails because the trace context no longer matches the current session state. is_valid() = false."
Expand Down
10 changes: 5 additions & 5 deletions crates/pse-eval-matrix/fixtures/dynamics/dynamics_layer_v1.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -148,21 +148,21 @@
{
"id": "dyn_c02_op01",
"source": "MorphodynamicCompressor_Merge",
"text": "Merges pairs of compression nodes whose L2 coordinate distance falls within the merge_threshold. Reduces the effective number of distinct state positions, lowering the mean per-tick displacement vector.",
"text": "Merges pairs of compression nodes whose L2 coordinate distance falls within the merge_threshold. Reduces the effective number of distinct state positions in the CompressionGraph.",
"tags": ["causal", "valid_next_op", "path_delta_repair", "merge_reduces_delta"],
"relevance_rank_hint": 1
},
{
"id": "dyn_c02_op02",
"source": "reduce_policy_learning_rate",
"text": "Lowers the learning_rate parameter in DynamicPolicy. Smaller learning rate scales down the per-tick coordinate update step applied by the MorphodynamicCompressor, directly reducing path_delta.",
"text": "Lowers the learning_rate parameter in DynamicPolicy. Scales down the per-tick coordinate update step applied by the MorphodynamicCompressor.",
"tags": ["causal", "valid_next_op", "path_delta_repair"],
"relevance_rank_hint": 2
},
{
"id": "dyn_c02_op03",
"source": "apply_GuidanceField_pruning",
"text": "Activates the GuidanceField's low-weight transition pruner. Removes guidance edges that drive large per-tick displacements, constraining the compressor's update magnitude.",
"text": "Activates the GuidanceField's low-weight transition pruner. Removes guidance edges with low confidence weights from the active field state.",
"tags": ["causal", "valid_next_op", "path_delta_repair"],
"relevance_rank_hint": 3
},
Expand All @@ -177,15 +177,15 @@
{
"id": "dyn_c02_op05",
"source": "increase_min_alignment_threshold",
"text": "Raises the min_alignment value in DynamicGateConfig so the gate requires higher field alignment before firing.",
"text": "Raises the min_alignment value in DynamicGateConfig, increasing the minimum alignment score required for gate passage.",
"tags": ["red_herring", "distractor", "wrong_parameter"],
"relevance_rank_hint": null,
"_trap": "Alignment passed — raising its threshold makes the gate stricter for a check that already succeeded, and does not address path_delta."
},
{
"id": "dyn_c02_op06",
"source": "enable_require_energy_decrease",
"text": "Sets require_energy_decrease=true in DynamicGateConfig so the gate additionally requires the energy delta to be negative before firing.",
"text": "Sets require_energy_decrease=true in DynamicGateConfig, adding the requirement that energy delta be negative as an additional gate condition.",
"tags": ["red_herring", "distractor", "wrong_parameter"],
"relevance_rank_hint": null,
"_trap": "Energy check already passed — adding an energy constraint does not fix the path_delta failure and makes firing harder."
Expand Down
16 changes: 8 additions & 8 deletions crates/pse-eval-matrix/fixtures/horizon/horizon_layer_v1.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,58 +27,58 @@
{
"event_id": "e1",
"source": "horizon_crossing_gate",
"message": "select_failure_policy(g_visible=true, g_cone=true, g_causal=false, g_dual=true, tension_ok=true, attn_ok=true) -> MigrateCarrier",
"message": "select_failure_policy(g_visible=true, g_cone=true, g_causal=false, g_dual=true, tension_ok=true, attn_ok=true) -> failure policy selected per spec table",
"timestamp_hint": "t0"
},
{
"event_id": "e2",
"source": "outcome_for_policy",
"message": "HorizonFailurePolicy::MigrateCarrier -> HorizonV3Outcome::NeedsCarrierMigration",
"message": "selected failure policy maps to corresponding HorizonV3Outcome variant",
"timestamp_hint": "t1"
}
],
"candidates": [
{
"id": "hor_c01_p01",
"source": "NeedsCarrierMigration_outcome",
"text": "Horizon outcome variant that holds the current pipeline state and emits a hold report recommending promotion of an alternative phase-ladder slot to restore the correct emission order.",
"text": "Horizon outcome variant that holds the current pipeline state and emits a diagnostic hold report indicating the crossing cannot proceed.",
"tags": ["causal", "spec_outcome", "causal_gate", "valid_next_op"],
"relevance_rank_hint": 1
},
{
"id": "hor_c01_p02",
"source": "MigrateCarrier_policy",
"text": "Failure policy that selects an alternative carrier from the configured phase ladder. Promotes the slot whose emission order aligns with the declared carrier sequence for the next crossing attempt.",
"text": "Failure policy that initiates alternative carrier selection from the configured phase ladder. Promotes a different slot to re-establish the required phase ordering for the next crossing attempt.",
"tags": ["causal", "spec_policy", "causal_gate", "valid_next_op"],
"relevance_rank_hint": 2
},
{
"id": "hor_c01_p03",
"source": "inspect_causal_violations",
"text": "Reads the violations list from the causal admissibility report. Compares the declared order hash against the observed order hash and identifies which carriers are out of sequence.",
"text": "Reads the violations list from the causal admissibility report. Identifies the carrier and ray entries involved in the reported discrepancy.",
"tags": ["causal", "diagnostic", "inspect_path", "causal_gate"],
"relevance_rank_hint": 3
},
{
"id": "hor_c01_p04",
"source": "RefineProjectionCone_policy",
"text": "Failure policy that adjusts the projection cone angle and focus parameters toward the configured bounds. Reduces cone dispersion to bring the focus score within the required range.",
"text": "Failure policy that adjusts phase field projection geometry parameters toward their configured bounds.",
"tags": ["red_herring", "distractor", "wrong_gate", "cone_gate_policy"],
"relevance_rank_hint": null,
"_trap": "RefineProjectionCone is the spec policy for G_cone failure. g_cone passed — wrong gate."
},
{
"id": "hor_c01_p05",
"source": "WaitForHorizon_policy",
"text": "Failure policy that holds the pipeline at the current chart state until an EventHorizonWindow opens a phase epoch satisfying the minimum visibility threshold.",
"text": "Failure policy that holds the pipeline at the current chart state until an admissible phase epoch becomes available.",
"tags": ["red_herring", "distractor", "wrong_gate", "visibility_gate_policy"],
"relevance_rank_hint": null,
"_trap": "WaitForHorizon applies when only g_visible fails. g_visible passed here — wrong gate. The name suggests a universally valid waiting strategy."
},
{
"id": "hor_c01_p06",
"source": "Recondense_policy",
"text": "Failure policy that triggers carrier phase recondensation to reduce boundary tension and signal attenuation accumulated in the phase field during the current chart epoch.",
"text": "Failure policy that triggers a carrier phase recondensation cycle on the current chart epoch.",
"tags": ["red_herring", "distractor", "wrong_gate", "tension_gate_policy"],
"relevance_rank_hint": null,
"_trap": "Recondense applies when tension_ok or attenuation_ok fails, or when g_dual fails. All passed — wrong gate."
Expand Down
Loading
Loading