Teleon Control Plane · `/app/queues`

Teleon enforces domain-specific atomic invariants at the database boundary across 320 single-verb queues, 61 short control aliases (aq01..aq61), and 760 domain shards: source discovery & 4-stage qualification (aq22..aq26), successor drift adaptation (aq32), reversible undo shadow pilots (aq34), Pareto non-regression (aq27), ratcheted hill-climbing, externality ideation, and beneficiary value attribution. Looking for durable versioned records? Open Persistent Stores (/app/stores).

6-Core Overview Atomic Queues (320)Persistent Stores (278)Muscle Memory & Fuzzy Recall File Vault Feed Streams Execution Router Orientation

All 78 Dedicated Atomic Work Queue Categories

aq09 · hill_climbing

Ratcheted Hill-Climbing & Pareto Frontier

Invariant: Deduplicates mutation_signature (409 on repeat) and only promotes incumbent when candidate_score > incumbent_score and guardrails_passed === true; suboptimal runs record negative knowledge.

Prevents parallel agents from overwriting improvements, regressing secondary guardrails, or re-testing known dead ends.

aq10 · externality_ideation

Externality Measurement & Second-Order Ideation

Invariant: Quantifies positive and negative 1st/2nd/3rd-order spillovers and atomically spawns a linked ideate_internalization work item whenever spillover crosses threshold.

Stops narrow Goodhart optimization from creating unpriced technical debt, UX harm, security risk, or missed positive-sum byproducts.

aq11 · value_attribution

Beneficiary & Value Attribution ('Who Actually Benefits?')

Invariant: Requires a normalized (sum = 1.0) Beneficiary Incidence Ledger naming concrete cohorts, evidence kind, and cost_bearer_cohort; flags incidence asymmetry when cost-bearers capture <20% of value.

Eliminates busywork theater by proving who captures real value vs. who bears the cognitive, financial, or operational cost.

aq12 · adversarial_falsification

Adversarial Falsification & Pre-Mortem Red-Teaming

Invariant: Enforces strict actor non-correlation at claim time (worker_id !== author_agent_id -> 409 on self-claim).

Prevents confirmation bias and rubber-stamp self-grading before high-impact releases.

aq13 · opportunity_cost

Counterfactual & Opportunity-Cost Gating

Invariant: Evaluates proposed value against mandatory zero_call_baseline_value and next_best_alternative_value before authorizing downstream spend.

Prevents spending LLM tokens or engineering weeks on tasks beaten by a simple deterministic script or off-the-shelf option.

aq14 · context_decomposition

Submodular Context-Slicing & Leaf Contract Merge

Invariant: Splits large epics into <=4k-token leaf leases with typed input/output contracts while keeping the parent vertex parked until leaf merge.

Allows smaller, cheaper models to execute large overnight tasks reliably without context window degradation.

aq15 · fragile_decay_reverify

Fragile-Fact Half-Life & Delta Re-Verification

Invariant: Schedules items by volatility half-life and emits downstream events only when content hash or benchmark rank mutates.

Keeps fast-rotting benchmarks, model releases, pricing, and deprecations accurate without redundant polling.

aq16 · quarantine_qualification

Untrusted Quarantine & Multi-Gate Qualification

Invariant: Requires full bitmask passage (secret scan, license check, deterministic test, schema validation) inside staging before catalog promotion.

Safely ingests donated public-good compute outputs, working-directory files, and 3D/media assets.

aq17 · blind_consensus

Blind K-Agent Consensus & Calibration

Invariant: Seals independent evaluations until K distinct workers commit verdicts, preventing herd anchoring.

Produces calibrated inter-rater reliability scores on ambiguous architectural or public-good evaluations.

aq18 · bottleneck_toil_mining

Human Relay & Organizational Toil Mining

Invariant: Ranks recurring manual copy-paste/approval loops by weekly_hours_lost * automation_feasibility.

Identifies where executives and engineers are acting as 'meat proxies' between disconnected systems.

aq19 · compatibility_matrix

Combinatorial Role x Harness Compatibility Verification

Invariant: Leases exact (role, harness, model_tier, skill_bundle) cells and records signed co-working verification receipts.

Guarantees that skills, tools, dependencies, and model configs actually work together before shipping an agent kit.

aq20 · negative_knowledge

Dead-End Pruning & Negative-Knowledge Synthesis

Invariant: Drains failed or suboptimal leases into compact 'Do Not Repeat' constraint cards indexed by task signature.

Turns failed experiments into reusable institutional memory so future agents never pay twice for the same mistake.

aq21 · muscle_memory_induction

Situation Fingerprinting, Semantic Blocking & Muscle-Memory Crystallization

Invariant: Computes 4-path SituationFingerprints (skeleton_hash + blk_dom/blk_io/blk_err/blk_lsh + effect_mask), blocks INFRASTRUCTURE_INVALID blips from semantic learning, promotes N>=3 independent verified episodes into 0-call deterministic reflexes, and trips a circuit-breaker on postcondition drift.

Graduates repeated Agent-led (1+ calls) and Hybrid workflows into sub-millisecond 0-model-call procedural muscle memory without O(N^2) vector scans or brittle exact-string caches.

aq22 · source_discovery

Source Discovery & Canonical URI Deduplication (find_sources)

Invariant: Deduplicates canonical source_uri (409 on repeat), registers the raw drop (Google Drive folder, Dropbox link, GitHub release, S3 bucket, EDGAR feed, or HTTP directory) in teleon_mt_source_registry, and atomically spawns a qualify_source task in aq23.

Turns ad-hoc folder links, APIs, and public data drops into tracked intake leases without duplicate crawling.

aq23 · qualify_source

Source Access, License, Safety & Modality Qualification

Invariant: Verifies reachability, access_mode, SPDX/terms compatibility, zero PII/harm vetoes, and detected file modalities; quarantines vetoed sources and atomically spawns BOTH qualify_searchability (aq24) and qualify_importance (aq25) when clean.

Prevents broken links, proprietary/unsafe drops, or PII leaks from entering Persistent Substrate Stores.

aq24 · qualify_searchability

Searchability, SKOS L0 Scent, IOPE & 8-Space Indexing Qualification

Invariant: Validates <=50-token SKOS L0 scent card, typed IOPE ports, 4-tier Roaring bitmaps (B_85, B_60, B_NEG), and modality-matched similarity space; computes searchability_score_0_1 and spawns OCR/ASR indexing in aq14 if < 0.60.

Guarantees every stored code module, contract, image, video, 3D scene, or dataset can actually be found and retrieved in <1 ms.

aq25 · qualify_importance

Expected Value of Information (EVOI) & Importance Tiering

Invariant: Computes non-compensatory importance_score_0_1 from expected_evoi_bits_per_100tok, taedri_novelty_delta, beneficiary_reach_0_1, and decision_criticality_0_1; assigns CRITICAL | HIGH | STANDARD | ARCHIVE_COLD | PRUNE and spawns aq26 taxonomy promotion when >= STANDARD.

Stops storage bloat and context dilution by separating high-leverage decision substrates from low-signal noise.

aq26 · qualify_categorization_taxonomy

Store S01..S278 Taxonomy Assignment & Catalog Promotion

Invariant: Binds qualified items to canonical Persistent Store IDs (S01..S278), store_family, modality, and Git-blob SWHID, committing verified Universal Substrate Envelopes into teleon_mt_substrate_catalog.

Bridges ephemeral qualification verbs (Atomic Queues) into versioned, content-addressed reusable nouns (Persistent Stores).

aq27 · pareto_non_regression_gate

Multi-Model Pareto Non-Regression & Mutation-Kill Gate

Invariant: Evaluates candidate Contract Cards and skills across multiple model families plus mutmut_kill_score >= 0.80; blocks promotion whenever min(per_model_deltas) < 0.

Prevents shipping instructions or skills that lift one model while silently regressing another model family.

aq28 · qualification_before_injection

Qualification-Before-Injection (<=2,048B Pre-Turn Card Gate)

Invariant: Bypasses injection when baseline_pass_prob >= 0.98 and ambiguity_entropy < 0.15, blocks cards > 2,048 bytes, and routes compact L1 Contract Cards strictly pre-turn.

Eliminates +62% latency bloat on already-saturated tasks and avoids mid-turn context disruption.

aq29 · consideration_set_foraging

Information-Scent & Consideration-Set Pruning (5,000 -> <=5)

Invariant: Applies Tversky B_NEG hard aspect vetoes followed by Pirolli/Card Information Scent (EVOI per token) to prune thousands of store candidates down to <=5 items.

Matches human and agent bounded rationality so workers compare 3-5 qualified options instead of drowning in 50 noisy hits.

aq30 · temporal_decision_review

Temporal Evidence Decay & Explicit Review-Trigger Audit

Invariant: Applies exponential staleness decay exp(-lambda * age_days) to Beta (alpha, beta) evidence posteriors and fires deterministic reopen leases when review_trigger predicates trip.

Re-evaluates past architectural, benchmark, or vendor decisions automatically when pricing, latency, or release facts shift.

aq31 · rux_harness_benchmark_audit

Fixed-Harness Anti-Benchmaxing & Capability Audit

Invariant: Enforces identical tool/prompt scaffold parity across models on uncontaminated holdouts and flags scaffold inflation whenever retry budgets or custom wrappers diverge.

Separates genuine model or substrate lift from benchmark-specific scaffold gaming.

aq32 · successor_drift_adaptation

Constructor / Checker / Successor Drift Adaptation & Amortization Gate

Invariant: Separates Constructor, Checker, and Successor affine-drift bracketing; compares retained_calls vs rebuild_calls, conventional_cache_calls, and direct_expert_calls over expected_reuse_count, emitting PROMOTED_SUCCESSOR_BRACKET, KEEP_PRESENT_DESIGN_CACHE_SUFFICIENT, or HALT_TRANSFER_UNAMORTIZED_OR_OUT_OF_BRACKET while keeping historical_evidence_swhid, qualification_state, and production_authority_scope strictly separate.

Turns completed missions into reusable operators for shifted conditions while knowing measurably when transfer saves calls (-12.69% vs rebuild), ties a cache, loses (+92 calls on unamortized one-offs), or must stop.

aq33 · problem_formulation_reframing

Problem-Formulation Challenge & Zero-Call Algebraic Reframing

Invariant: Audits whether an agentic workflow prompt masks an exact deterministic algebraic, SQL-join, or set-reconciliation operator; emits REFRAMED_TO_ZERO_CALL_OPERATOR into S246 when algebraic_reducibility_verified === true and estimated_call_reduction_pct >= 50.

Challenges poorly framed tasks before burning expensive LLM loops when a deterministic operator solves the true underlying need.

aq34 · reversible_undo_shadow_pilot

Session-Tagged Reversible Undo, Import-Closure & No-Send Shadow Pilot Gate

Invariant: Requires non-empty session_tag, inverse patch undo_bundle_swhid, import_graph_closed === true (zero omitted packaged dependencies), target_refs_fresh === true (live DOM/API bindings unmutated mid-batch), and no_send_shadow_verified === true before side effects.

Prevents multi-agent workspace clobbering during rollback, ERR_MODULE_NOT_FOUND bundle crashes, stale mid-batch UI mutations, and unverified external sends.

aq35 · abstention_proportionality_gate

Proportionality, Selective Escalation & Auditable Abstention Ledger

Invariant: Evaluates evidence_score against policy_threshold before intrusive inspection, camera escalation, or high-impact alerts, committing an immutable docket to Store S248 recording both why the system looked and why it held back.

Eliminates surveillance creep, alert fatigue, and unlogged silent drops across civic safety, compliance, and trust-sensitive workflows.

aq36 · world_state_canon_consistency

Persistent World-State, Entity-Timeline & Canon Invariant Verification

Invariant: Checks multi-session state transitions against entity-timeline causal order, lifecycle/ownership invariants, and lore axioms in Store S249, blocking canon contradictions at commit time.

Keeps persistent interactive worlds, simulations, and long-horizon stateful narratives coherent across hundreds of sessions.

aq37 · accessibility_sensory_verification

Screen-Reader, Non-Visual Navigation & Sensory Accessibility Gate

Invariant: Verifies semantic accessibility tree completeness, deterministic focus order, ARIA live-region announcements, tactile/audio fallback cues, and visual_only_affordances === 0.

Guarantees blind, low-vision, and eyes-free users receive complete non-visual operational parity instead of visual-only UI dead ends.

aq38 · sensor_calibration_feasibility

Physical Sensor Calibration, Artifact Rejection & BOM Feasibility Gate

Invariant: Enforces physical reference brackets (snr_db >= min_snr_db, 50/60Hz notch and ocular artifact rejection, spatial fiducial/ArUco scale lock, and unit_bom_usd <= max_bom_usd) prior to downstream model inference.

Stops models from guessing over uncalibrated EEG/BCI signals, scale-free AR plant cameras, or hardware prototypes that violate unit-cost ceilings.

aq39 · ephemeral_sandbox_reconciliation

Zero-Password Ephemeral Sandbox & Two-List Ledger Reconciliation

Invariant: Rejects shared plaintext demo credentials (shared_plaintext_credentials === false), verifies TTL-scoped trial token isolation, and executes deterministic two-list double-entry shipment/payment reconciliation with source-linked exception grouping.

Replaces leaked sandbox passwords in public channels and catches partial-failure logistics and fintech settlement mismatches deterministically.

aq40 · audio_reactive_keyframe_sync

Real-Time Audio Beat/Onset Phase & 60-FPS Frame-Budget Sync Gate

Invariant: Verifies beat/onset phase drift abs_onset_error_ms <= 15 and render p99_frame_ms <= 16.67 across audio stems and reactive shader keyframes.

Provides deterministic timing and frame-budget verification for real-time audio-reactive visuals, live stages, and synchronized captions.

aq41 · verified_hands_on_mission_check

Fixture-Verified Hands-On Micro-Mission & Fact-vs-Inference Gate

Invariant: Grades hands-on task outputs against golden fixtures within a bounded step_budget (default <= 8) while requiring strict separation of cited observed_facts from inferences.

Replaces passive quizzes and unbounded codebase wandering with reproducible, citation-backed skill verification.

aq01 · standard_fenced

Monotonic Lease-Epoch CAS Dispatch

Invariant: Increments lease_epoch atomically (meta.changes === 1) on claim and rejects stale worker heartbeats or completions with 409 fencing_epoch_conflict.

Eliminates split-brain duplicate commits across concurrent distributed worker fleets.

aq229_qualification_before_injection · pre_turn_contract_routing

Pre-Turn <=2,048B Diagnostic Contract Routing

Invariant: Routes compact <=2,048-byte L1 Contract Cards strictly prior to Turn 1 when task ambiguity or root-cause diagnostic risk is positive.

Lifts multi-step investigative pass rates (0% -> 87.5% on diagnostic benchmarks) without mid-turn trajectory disruption.

aq230_consideration_set_foraging · information_scent_evsi

Pirolli & Card Information Scent & EVSI Patch-Leaving

Invariant: Scores candidate substrate patches by Expected Value of Sample Information (bits per 100 tokens) over <=50-token L0 cards before loading full payloads.

Prevents blind full-document context stuffing across large folders and repositories.

aq260_elimination_by_aspects_veto · elimination_by_aspects_veto

Tversky Elimination-by-Aspects Roaring B_NEG Hard Veto

Invariant: Applies <0.15 ms bitwise ANDNOT against B_NEG facets prior to similarity ranking so unsafe licenses, PII, or withdrawn crypto never enter consideration sets.

Guarantees high semantic similarity can never override hard compliance or security boundaries.

aq226_muscle_memory_induction · recognition_primed_preset_promotion

Klein Recognition-Primed DMA Preset Promotion

Invariant: Promotes recurring verified decision configurations into 1-click presets only after N>=3 clean episodes with failure_streak === 0.

Turns proven incident and engineering playbooks into sub-millisecond reusable presets without single-shot luck pollution.

aq242_primitive_extraction · taedri_primitive_extraction

Mutation-Tested Reusable Code Primitive Extraction

Invariant: Verifies stdlib-first Python, TypeScript, and SQL primitives with mutmut kill score >= 0.82, typed IOPE ports, and Git-blob SWHID.

Stops agents from rewriting fragile parsers and helpers from scratch on every task.

aq243_wasm_deterministic_compilation · wasm_deterministic_compilation

Fuel-Bounded Deterministic Wasmtime .cwasm Compilation

Invariant: Compiles pure primitives with canonical NaNs, zero ambient clock/net WASI imports, and <=5,000,000 instruction fuel ceilings.

Enables sandboxed, bit-exact 0-model-call execution across heterogeneous x86_64 and ARM64 workers.

aq244_lean4_formal_proof_check · lean4_formal_proof_check

Lean 4 Machine-Checked Protocol Invariant Verification

Invariant: Requires lake build exit code 0 with zero sorry or unapproved axiom declarations across formal state-machine and conservation specs.

Eliminates subtle epoch-wrap, CAS race, and basis-point rounding bugs in critical state machines.

aq245_gltf_openusd_scene_validation · gltf_openusd_scene_validation

glTF 2.0 / OpenUSD Scene IR & WebGPU Budget Gate

Invariant: Validates gltf-validator schema compliance, KTX2 texture compression, explicit Y-up meter metadata, and <=180 draw calls.

Prevents unoptimized textures, broken axis conventions, and mobile WebGPU crashes in 3D pipelines.

aq246_cad_parametric_zernike_mesh · cad_parametric_zernike_mesh

Parametric STEP Watertight B-Rep & 3D Zernike Indexing

Invariant: Verifies positive watertight solid volume via BRepCheck_Analyzer and computes order-12 3D Zernike descriptors for rotation-invariant search.

Catches non-manifold open shells before CNC/3D printing and enables geometric similarity search.

aq247_video_treatment_blueprint_render · video_treatment_blueprint_render

Multi-Treatment Editorial Motion & Chart Render Verification

Invariant: Enforces 8% platform title-safe margins, zero-baseline bar charts, deterministic headless frame export, and webhook idempotency.

Prevents clipped mobile overlays, misleading truncated chart axes, and duplicate social/email publishes.

aq248_photo_wcag_palette_audit · photo_wcag_palette_audit

WCAG 2.2 AAA Contrast Palette & Pure Fourier SVG Audit

Invariant: Requires >=7.0:1 normal text contrast, >=4.5:1 UI border contrast, CVD series distinguishability, and bounded-precision raster-free SVG paths.

Eliminates low-contrast UI labels and bloated raster-embedded brand/diagram assets.

aq249_audio_chromagram_asr_stem · audio_chromagram_asr_stem

Clean-Room Foley, -14 LUFS Loudness & ASR Alignment Gate

Invariant: Verifies -14 LUFS loudness, <=-1.0 dBTP true peak, zero copyrighted sample matches, and forced-alignment ASR transcript normalization.

Prevents inter-sample codec clipping, copyright contamination, and casing-induced WER false alarms.

aq250_legal_worker_spdx_compliance · legal_ofw_spdx_compliance

Worker-Protection Contract, SaaS DPA & SPDX License Gate

Invariant: Hard-vetoes placement fee deductions or passport retention, enforces <=72h DPA breach notice clauses, and checks transitive SPDX compatibility.

Protects vulnerable workers from fee traps and blocks copyleft or vendor-training clause leaks.

aq251_edgar_xbrl_catalyst_extraction · edgar_xbrl_catalyst_extraction

SEC EDGAR 8-K/10-Q XBRL Tie-Out & Catalyst Audit

Invariant: Pins exact accession numbers and comparative periods, verifies $1 balance-sheet/cash-flow tie-outs, and separates energized MW from LOIs.

Stops financial hallucination across split-adjusted share counts, covenants, and infrastructure capex filings.

aq252_municipal_311_entity_dedup · municipal_311_splink_dedup

Municipal Socrata 311 Normalization & Probabilistic Linkage

Invariant: Normalizes SODA timestamps/coordinates, preserves USPS apartment/suite designators, and verifies >=0.94 precision/recall on labeled holdouts.

Prevents duplicate civic dispatch orders while avoiding false household merges across multi-unit buildings.

aq201_tokenize_cleanroom_ff1_radix_0call · nist_ff1_cleanroom_privacy

NIST SP 800-38G Rev.1 FF1-Only Tokenization & 8-Gram Guard

Invariant: Enforces AES/HMAC-FF1 format-preserving tokenization (radix^n >= 1,000,000), hard-refuses withdrawn FF3/FF3-1, and checks 8-gram originality.

Protects regulated PII and clean-room IP boundaries without breaking fixed-width downstream schemas.

aq253_sdg_humanitarian_commons_audit · sdg_humanitarian_commons_audit

17-SDG Zero-Fee Humanitarian & Clinic Commons Gate

Invariant: Enforces 0-bps Ring-3 fee waiver, permissive CC0/MIT/CC-BY-4.0 open licensing, client-side PII redaction, and offline clinic usability.

Delivers safe, lock-in-free tools to NGOs, rural clinics, and anti-human-trafficking responders.

aq254_exec_delegation_card_synthesis · exec_delegation_card_synthesis

<=150-Token Executive Delegation & Boundary Card Synthesis

Invariant: Caps delegation contracts at <=150 tokens with explicit owner, objective rubric, budget cap, and Means-vs-Ends escalation predicates.

Replaces multi-day Slack/email relay loops with crisp, verifiable delegation contracts.

aq255_overnight_leaf_pack_assembly · overnight_leaf_pack_assembly

Mode B <=4,000-Token Overnight Leaf Context Assembly

Invariant: Caps self-contained leaf context packs at <=4,000 tokens paired with a deterministic local exit-0 test command.

Keeps small 8B-32B local and cloud overnight workers accurate without KV-cache overflow.

aq256_eight_file_kit_packaging · eight_file_kit_packaging

Deterministic 8-File Working-Directory Kit Packaging

Invariant: Verifies presence, schema validity, SHA-256 manifest, and executable verifier pass across all 8 canonical working-directory kit files.

Ships drop-in repository skills and hooks that run reliably across Claude Code, Codex, OpenCode, Cursor, and Pi.

aq257_artifact_nutrition_holdout_lift · artifact_nutrition_holdout_lift

Paired Holdout-Lift Artifact Nutrition Labeling

Invariant: Measures paired (with - without) task pass delta, token delta, and sample count N on held-out tasks unseen during skill authoring.

Replaces vanity star counts and self-graded prompts with empirical trust nutrition labels.

aq227_fuzzy_cognitive_step_clustering · fuzzy_cognitive_step_clustering

v1.4 5-Signal Fuzzy Cognitive-Step SimHash64 Indexing

Invariant: Indexes phase(16), obstacle(12), concept_lattice(24), char 3-grams, and 8-band SimHash64 to surface what_worked and what_failed_avoid in <1.5 ms.

Recalls structurally analogous reasoning steps even when domain vocabulary or spelling differs completely.

aq258_confusion_sibling_disambiguation · confusion_sibling_disambiguation

SKOS Sibling Confusion-Boundary Disambiguation

Invariant: Identifies co-retrieved false-positive sibling substrates and attaches contrastive When-A-not-B rules and negative facet tags.

Sharpens retrieval precision when multiple tools or schemas share overlapping domain keywords.

aq259_spare_token_pareto_composition · spare_token_pareto_composition

Off-Peak Spare-Token Kleisli Pareto Composition

Invariant: Composes 0..32 pure typed schema bridges during off-peak token windows and promotes only SWHID outputs beating single-call cost/latency.

Converts idle off-peak token quotas into sub-millisecond 0-call production bridges.

aq42 · actor_intent_deduplication

Actor Observation Deduplication & Intent Parsing

Invariant: Dedicated actor table join; merges repeated chat introductions into 1 canonical Actor with N linked observations.