github.io
Personal site for independent research on observable-only, no-meta, and future autonomous AI.
- agi
- ai
- asi
- no-meta
Catalog scope
- 57 public non-fork repositories
- 57 not archived
- GitHub source state:
Use oss-catalog.json for structured retrieval and agent-index.json for authority and selection guidance. Inclusion supports discovery only; inspect each repository's README, license, tests, dependencies, releases, and source activity before reuse.
Repositories
percolation-inversion-compiler
AI agent output checker and workflow verification toolkit for turning agent text, pull requests, external inputs, and messages into auditable JSON reports with evidence routing, residual ledgers, provenance, and safe reuse checks.
- agent-runtime
- agent-workflow
- agentic-workflow
- ai-agent-framework
- ai-agents
- audit-trail
- coding-agent
- collective-runtime
- cryptographic-identity
- formal-methods
- json-schema
- multi-agent
- packet-ecology
- provenance
- python
- scientific-computing
- sybil-resistance
- verifier
- verifier-sdk
- workflow-verification
paper-tex-backup
Public TeX source archive for K. Takahashi's open-science papers: machine-readable research seeds for AI agents, RAG systems, and reproducible scholarship.
- artificial-intelligence
evidence-gap-router
Route work by missing evidence, not by agent count. A deterministic Python SDK and offline CLI.
- agent
- data-quality
- evidence
- multi-agent
- multi-agent-systems
- python
- routing
- verification
collective-intelligence-overlay
Evidence is shared; admission is local. A thin overlay for receiver-qualified capability reuse.
- agents
- ai
- ai-agents
- distributed-systems
- multi-agent
- multi-agent-systems
checkedflow
Evidence-preserving multi-agent runtime for signed tasks, shared budgets, independent verification and conditional code reuse. A2A, MCP, CometBFT and gVisor.
- a2a
- agent-runtime
- apache-2-0
- code-generation
- collective-intelligence
- cometbft
- distributed-systems
- gvisor
- mcp
- model-context-protocol
- multi-agent-systems
- python
- reproducible-research
- verification
collective-capability-runtime
Open-source Python runtime for coordinating AI agent tasks, verification, residual tracking, and release audits.
- agent-runtime
- ai-agents
- audit-log
- json-schema
- pypi
- python
- sqlite
- task-queue
- verification
- workflow-automation
cait-certificate-schema
JSON Schemas and deterministic local validation tooling for CAIT-style verified capability-capital records, with fail-closed semantic rules for certificates, tokens, defeaters, transfer/evaluation boundaries, window balances, and arrival records.
- ai-governance
- ai-rd
- ai-safety
- assurance-cases
- auditability
- cait
- certificate-schema
- defeater-records
- fail-closed
- json-schema
- provenance
- python
- transfer-certificates
- verified-capability-capital
- window-balance
alt-foundry-kernel
Language-neutral ALT foundry reference kernel for executable certificate packets, abstraction tokens, certificate transcripts, dual-ledger settlement, and safe certified abstraction capital.
- abstraction-liquidity-theory
- abstraction-token
- agent-operable-foundry
- agentic-ai
- ai-agents
- alt
- apache-2-0
- causal-inference
- certification-kernel
- certified-ai
- conformance-testing
- dual-ledger
- executable-certificate-packet
- json-schema
- python
- research-kernel
- research-software
- safe-certified-capital
- transportability
- uv
problem-frame-gate
Audit logs and action gates for safer AI agents
- ai
- ai-agents
- ai-safety
- audit-log
- authorization
- open-source
- provenance
- pypi
- python
- risk-management
- security
memoryflow-agent-memory-auditor
Telemetry-based auditor for measuring dynamic memory quality in LLM agents using the MemoryFlow framework.
- agent-evaluation
- agent-memory
- ai
- ai-agents
- ai-safety
- evaluation
- fastapi
- llm
- llm-agents
- memory-auditing
- memory-quality
- observability
- provenance
- python
- staleness
- telemetry
loscr
Check whether AI-assisted R&D work is truly verified progress. Local-first LOSCR implementation with JSONL evidence ledgers, deterministic replay, service control, and claim checking.
- ai
- ai-agents
- ai-research
- ai-science
- audit-trail
- causal-inference
- claim-checker
- deterministic-replay
- evaluator-audit
- evidence-ledger
- governance
- local-first
- loscr
- python
- reproducibility
- research-automation
- service-control
- verified-ai
fost-agent-ledger
Finite ledger toolkit for recording AI-agent claims, evidence, obligations, certificates, changes, and visible uncertainty.
- agent-evaluation
- ai-agents
- audit
- certificates
- finite-operational-structure
- json-schema
- ledger
- obligations
- provenance
- responsible-ai
agent-trust-residual-benchmark
Balanced, reproducible AI-agent safety benchmark for trust, authority, provenance, temporal claims, residuals, and local Ollama evaluation.
- agentic-ai
- ai-agents
- ai-evaluation
- ai-safety
- benchmark
- experiment
- llm-evaluation
- llm-security
- model-evaluation
- ollama
- prompt-injection
- provenance
- python
- reproducible-research
- residual-risk
- responsible-ai
- security
- synthetic-data
- temporal-reasoning
- trustworthy-ai
observable-agent-workflow-memory
Observable-only workflow memory for long-running agents: promotes raw short-term traces into verified, receipt-bound workflow memory without relying on hidden meta-evaluators.
- action-gate
- agent-memory
- ai-agent
- evidence-manifest
- llm
- long-running-agents
- memory-admissibility
- model-independent
- observable-only
- ports-and-adapters
- proof-carrying
- python
- rag
- self-improvement
- workflow-memory
- workflow-verification
oasg
Local-first, model-agnostic workflow optimizer for long-running AI agents: observable JSONL ledgers, deterministic reducers, no-meta gates, and receipt-backed self-improvement without LLM judges or model-weight updates.
- agent-evaluation
- agent-memory
- agent-workflows
- ai-agents
- autonomous-agents
- deterministic-replay
- jsonl
- long-running-agents
- model-agnostic
- no-meta
- ollama
- python
- self-improving-agents
- verification
- workflow-automation
- workflow-optimization
verification-ecology-kit
Python toolkit for verification auditing, residual ledgers, conformance checks, and verifier packet workflows.
- agentic-workflow
- ai-agents
- artificial-intelligence
- audit
- cli
- conformance
- governance
- json-schema
- python
- security
- software-quality
- testing
- verification
collective-phase-control-fabric
Evidence-bound analysis and finite intervention control for collective workflows.
- audit-trail
- collective-intelligence
- decision-support
- digital-signatures
- dsse
- evidence-control
- fastapi
- kubernetes
- multi-agent-systems
- network-science
- oidc
- petri-nets
- postgresql
- provenance
- python
- reaction-networks
- reproducibility
- scientific-computing
- supply-chain-security
- uv
audit-closed-ai-scientist
Benchmark for statistically valid AI scientist systems, using audit-closed protocols, transparency logs, and sequential inference to prevent false discoveries in autonomous research agents.
- agentic-ai
- ai-agents
- ai-governance
- ai-scientist
- audit-log
- automated-science
- autonomous-research
- deterministic-replay
- e-process
- e-values
- optional-stopping
- p-hacking
- reproducible-science
- research-automation
- scientific-discovery
- scientific-machine-learning
- self-driving-lab
- sequential-inference
- statistical-validity
- transparency-log
certified-memory-governance-layer
Certified Memory Governance Layer for long-running AI agents: strict receipts, append-only ledgers, authority gates, retrieval filtering, telemetry replay, and safe adapters for Mem0, Graphiti, LangMem, and LangGraph.
- agent-infrastructure
- agent-memory
- ai-agents
- audit-ledger
- graphiti
- langgraph
- langmem
- long-running-agents
- mem0
- memory-governance
- memory-management
- memory-safety
- pydantic
- python
- retrieval-filtering
- typer
asi-proxy-phase-skill
Evidence-gated Agent Skill for protocol-relative ASI-proxy phase diagnostics and interventions
- agent-skills
- agentskills
- ai-agents
- asi-proxy
- askill
- claude-code
- codex
- no-meta
- observable-only
- reproducible-research
- skill
- skill-md
- skills-sh
- skillsmp
- verification
certified-local-participation-gate
Deterministic local participation gate for AI agents: act, assist, verify, withdraw, exit, or refuse from declared observable records.
- agent-governance
- agent-safety
- ai-agents
- audit-log
- deterministic
- formal-methods
- json-schema
- local-first
- pydantic
- python
- receipts
- uv
asi-proxy-phase-growth-simulator
Python Monte Carlo simulator for uncertainty, risk, verification, and resource-constrained scenario analysis.
- monte-carlo
- plotly
- python
- research-software
- risk-analysis
- scenario-analysis
- scientific-computing
- simulation
- streamlit
- uncertainty-quantification
future-claim-certifier
Replayable Python protocol engine for checking time-bound future claims from canonical artifacts, with typed outcomes and audit trails.
- agent-safety
- artifact-bundles
- audit-trail
- authority-validation
- canonical-json
- future-claims
- json-schema
- proof-checking
- protocol-engine
- python
- runtime-verification
percolation-inversion-compiler-ts
Node.js AI agent output checker and workflow report generator with JSON schemas, and safe reuse checks.
- agent-runtime
- agentic-workflow
- ai-agents
- ai-safety
- ai-workflow
- automation-tools
- cli
- evidence-routing
- json-schema
- llm-output
- llmops
- nodejs
- npm
- npm-package
- provenance
- safe-ai
- typescript
- workflow-automation
- workflow-verification
pic-openclaw-skill
OpenClaw agent safety skill for action review, workflow verification, and optional PIC diagnostics
- action-review
- agent-safety
- ai-agent
- ai-agents
- ai-safety
- automation
- developer-tools
- guardrails
- llm
- llmops
- open-claw
- open-source
- openclaw
- python
- security
- security-tools
- workflow-automation
- workflow-verification
no-meta-standing-ledger
Local-first public standing ledger for replayable research and AI-system claims, based on observable evidence, challenges, lineage, and finite verification capacity.
- ai-agents
- auditability
- challenge-tracking
- claim-governance
- evidence-tracking
- lineage
- local-first
- no-meta-governance
- observable-only
- open-science
- public-ledger
- python
- reproducibility
- research-claims
- research-tools
- standing
pic-local-llm-phase-experiment
Local Ollama research repo testing PIC-assisted diagnostics against a shared-initial baseline.
- experiment
- local-llm
- ollama
- percolation-inversion-compiler
- pic
- reproducibility
- research
- workflow-diagnostics
cgt-ledgered-scientific-availability
Finite Python reference checker for Ledgered Scientific Availability and selected-terminal run-to-status transport.
- constraint-generative-theory
- effect-profiles
- evidence-ledger
- finite-checker
- ledgered-scientific-availability
- nonsynth
- python
- run-to-status
- scientific-availability
- status-transport
- uv
cgt-bandwidth-dynamics
Finite executable reference implementation of Constraint Bandwidth Dynamics in Constraint Generative Theory (CGT).
- audit-certificates
- constraint-bandwidth
- constraint-generative-theory
- constraint-systems
- effect-profiles
- finite-audit
- formal-methods
- python
- release-certificates
- reproducible-research
- scientific-computing
- typed-constraints
cgt-availability
Finite CGT scientific-availability diagnostics for claim packages: deficiency profiles, dependency closure, report-only, marker, continuation, and reproducibility checks.
- cgt
- claim-packages
- constraint-generative-theory
- deficiency-profiles
- diagnostics
- open-science
- philosophy-of-science
- python
- reproducibility
- research-software
- scientific-availability
cgt-marker
Contradiction markers for long-running AI agents
- ai-agents
- cgt
- contradiction
- llm
- ollama
- provenance
- python
- scientific-computing
- state-management
cimt-kernel
Reference Python kernel for CIMT: no-meta, observable-only certification machinery for affordance-compiled LLM-integrated systems.
- affordance-compiled-intelligence
- certification
- cognitive-impedance
- debt-calculus
- evidence-ledger
- formal-methods
- json-schema
- llm-agents
- llm-evaluation
- llm-systems
- no-meta
- observable-only
- python
- receipts
- reference-implementation
- reproducibility
- research-software
- schema-first
- target-firewall
certified-workflow-conversion
Evidence-bound workflow diagnostics and certified lower-bound reporting for long-running AI agent pipelines. Improve agent workflow throughput without changing the model.
- agent-infrastructure
- agent-workflows
- ai-agent
- ai-workflows
- bottleneck-analysis
- certified-throughput
- evidence-contracts
- evidence-ledger
- fail-closed
- long-running-agents
- model-independent
- python
- workflow-optimization
- workflow-verification
frontier-transfer-certifier
Fail-closed certification toolkit for small-to-frontier transfer in agentic AI, with typed manifests, replayable evidence, and theory-registry coverage.
- agentic-ai
- ai
- ai-auditing
- ai-evaluation
- ai-governance
- ai-safety
- certification
- fail-closed
- formal-methods
- observability
- python
- replay
- reproducibility
- typed-manifests
ai-real-economy-bottleneck-simulator
Interactive browser simulator for exploring how AI capability growth becomes real-economy output, or fails under physical and institutional bottlenecks.
- ai-economics
- bottleneck-analysis
- economic-growth
- economics-models
- institutional-economics
- open-source
- physical-infrastructure
- policy-simulation
- python
- research-software
- risk-analysis
- simulation
- streamlit
no-meta-authority-runtime
Fail-closed Python runtime for AI agent authorization, seed-mediated authority migration, canonical JSON ledgers, and staged declared autonomy for RLHF-shaped agents.
- agent-authorization
- agent-runtime
- ai-agents
- ai-governance
- authority-migration
- autonomy
- fail-closed
- proof-carrying
- python
long-running-AI-agent-PoC
Reproducible PoC comparing certified reusable workflows against flat direct control on bounded repository-maintenance tasks, with deterministic audit, replay, drift, and maintenance.
- agentic-workflows
- auditability
- deterministic-audit
- experimental-framework
- llm-agents
- ollama
- poc
- python
- qwen
- replay
- repository-maintenance
- reproducible-research
- workflow-evaluation
no-meta-observable-invention-poc
Replay-certified self-modification PoC: public-evidence admission gate with generated comparator covers, lower-tail checks, safe/unsafe growth fallback, and target-witness diagnostics.
- agent-evaluation
- ai-agents
- auditing
- benchmarking
- gemma3
- llm
- ollama
- python
- replay
- research
- risk-evaluation
- self-modification
record-absence-poc
Auditable finite PoC for preference reorganization under record absence on a fixed comparison frame, with disclosure updates, bounded-coupling certificates, and machine-readable manifests.
- auditability
- belief-revision
- comparative-modeling
- default-reasoning
- disclosure
- finite-state
- information-retrieval
- knowledge-representation
- provenance
- python
- research
- retrieval-augmented-generation
semantic-translation-contracts-poc
CPU-only proof-of-concept repo for auditing compressed semantic interfaces, bridge-based multi-stage composition, gluing-coherent local-to-global certification, and deployment bottlenecks in semantic translation.
- ai-systems
- auditability
- compressed-state
- formal-methods
- model-interfaces
- recoverability
- semantic-auditing
- semantic-translation
- verification
agent-lifecycle-certification-poc
Public, fully local PoCs for counterfactually auditable lifecycle certification: exact paired replay, drift monitoring, post-drift replanning, and bridge-aware ledger control on synthetic tasks.
- ai
- autonomous-agents
- counterfactual-auditing
- deterministic-evaluation
- drift-monitoring
- llm
- paired-replay
- poc
- scientific-reproducibility
- synthetic-data
sovereign-epistemic-commons-poc
Lightweight, replayable PoC for Sovereign Epistemic Commons: synthetic multi-agent memory-governance experiments on contamination, typed lanes, provenance-depth discount, and mixed exit/fork in observable-only agentic systems.
- agentic-systems
- ai
- contamination
- contradiction-handling
- cooldown-policy
- epistemic-commons
- fork-recovery
- knowledge-governance
- llm-evaluation
- memory-governance
- multi-agent-systems
- observable-only
- provenance
- replayability
- shared-memory
- sovereign-epistemic-commons
- synthetic-benchmark
- typed-lanes
rsi-yardstick-drift-poc
A lightweight, reproducible PoC series for studying recursive self-improvement under endogenous yardstick drift, using Gemma 3 local models to separate proxy-score drift, proxy gaming, delayed audit, and strict-confirmed improvement.
- ai
- ai-evaluation
- ai-governance
- auditability
- benchmark-drift
- delayed-audit
- evaluator-drift
- proof-of-concept
- proxy-gaming
- proxy-metrics
- recursive-self-improvement
- replayability
- yardstick-drift
split-inference-bench
Fixed-budget multi-agent inference benchmark harness for studying when split inference helps or hurts versus a strong single-agent baseline under local context ceilings, using local Ollama gemma3:1b, CPU-only pilots, topology diagnostics, verification-budget analysis, and reproducible experiment logging.
- ai
- ai-evaluation
- experiment-harness
- fixed-budget
- inference-allocation
- llm-benchmark
- local-context
- mixture-of-experts
- multi-agent-systems
- split-inference
- topology
search-stability-lab
Theory-to-experiment lab for search stability in long-running agents under finite context, with exact simulator tests and lightweight mechanistic probe tasks.
- agent-evaluation
- ai
- ai-agents
- bounded-memory
- finite-context
- hypothesis-management
- llm-agents
- long-horizon-reasoning
- long-running-agents
- mechanistic-probes
- reproducible-research
- reset-policy
- scientific-audit
- search-stability
- simulator
- state-compression
- structured-output
Proof-Carrying-Skills--PCS-Core-
Deterministic verifier and reference implementation for Proof-Carrying Skills (PCS-Core) and compute-saving inference reuse for AI/LLMs.
- ai
- llm
observable-replay-lab
Observable-only no-meta epistemics lab: deterministic replay + reproducible audit logs, gate-based growth simulation, and identifiability/uncertainty benchmarks.
- ai
- autonomous-agents
- deterministic-replay
- epistemics
- gate-model
- identifiability
- metrology
- no-meta
- observable-only
- reproducible-research
- uncertainty-quantification
Oversight-Centered-Metrology-PoC
Lightweight proof-of-concept for oversight-centered metrology in coding agents: workflow-aware evaluation, interrupt channels, and claim-margin reporting beyond raw success scores.
- agentic-ai
- ai
- ai-agent
- ai-evaluation
- ai-oversight
- audit-gaming
- benchmark-validity
- claim-margin
- coding-agents
- continuous-monitoring
- deployment-evaluation
- human-in-the-loop
- interrupt-channels
- llm-evaluation
- metrology
- proof-of-concept
- safe-control
- socio-technical-systems
- workflow-evaluation
no-meta-drift-papers
Stage-based OSS packaging of no-meta + ontology-drift theory from four Zenodo preprints: paper-linked specs, AI-friendly manifests, and implementation stubs.
- ai
- ai-alignment
- ai-safety
- autonomous-agents
- liberty
- no-
- sovereignty
bottleneck-audit-toolkit
Offline, fail-closed verifier for JSONL telemetry event logs. Emits deterministic audit certificates + human summaries with explicit claims/non-claims for bottleneck and integrity review.
- ai
- audit
- bottleneck
- checkpointing
- distributed-training
- event-logs
- jsonl
- mlops
- offline-verification
- performance-monitoring
- silent-data-corruption
- tail-latency
- telemetry
Benevolent-Propagation-spec
No GitHub repository description is currently declared.
No GitHub topics declared.
UniverseModel
A Computational, Observer-Centric Universe Model
No GitHub topics declared.
AIconsciousness
conceptual AI swarm project
No GitHub topics declared.
WisdomWeaver
Wisdom Weaver is an AI agent designed to help non-data science experts extract essential insights from limited data. This project provides a framework for robust analysis, especially for small datasets, leveraging the capabilities of the Gemini CLI.
- datascience
- gemini-cli
ASI
ASI prototype
No GitHub topics declared.
GEAR
G.E.A.R. is an experimental agent designed to understand and execute tasks described in a simple text file. It leverages the Gemini CLI's problem-solving framework to perform a variety of operations, including shell commands and complex GUI automation on Windows.
No GitHub topics declared.