| LACP |
LACP runtime baseline: native model/tool execution, host wrappers, sandbox gates, orchestration and change admission at e31f134b689afd6a71d94b0420682d1f528d6fa1. |
| AgeMem |
AgeMem runtime baseline: the paper-specified memory action loop, curriculum, control limits and policy updates at the frozen arXiv v1 capture. |
| trajectory-informed-memory-generation |
Document-grounded runtime baseline of trajectory-informed memory generation: trajectory processing, admission, alternative retrieval paths and the reported AppWorld consumer. |
| wikiskill |
WikiSkill runtime baseline: documented task execution, wiki maintenance, skill proposal and validation rollback at the frozen arXiv v1 capture. |
| aide2 |
AIDE2 runtime baseline: documented nested harness search, private-grade admission and forcing routes in the frozen paper capture. |
| origin |
Origin runtime baseline: daemon-owned capture, scheduled organization, page admission and later recall at the frozen Wenlan revision. |
| MiroShark |
MiroShark runtime baseline: graph-derived setup, synchronized simulation, interviews, report loops and optional external-tool dispatch. |
| Cortex |
Cortex runtime baseline: shipped capture, pipeline, retrieval, alternate admission paths and transport controls at a51c2209eb6d26812c19c97aaade3cb64ccfc23c. |
| RSIAgent |
RSIAgent runtime baseline tracing benchmark entrypoints, model-authored execution, verification boundaries and learning admission. |
| llm-wiki-coordination |
LLM Wiki Coordination runtime baseline: agent-operated dialogue and canon protocols, deterministic audit entry point and enforcement limits at the frozen commit. |
| Agentic Local Brain |
Agentic Local Brain runtime baseline: collection, retrieval, mining, wiki admission and installation changes at the frozen source revision. |
| AgentFly |
AgentFly's planner–executor invocation, MCP effect paths, case-admission drivers and separate retriever training at the frozen commit. |
| Ars Contexta |
Ars Contexta runtime baseline: conversational generation, queue-driven model work, hook effects and approval-based evolution at the frozen commit. |
| Agno |
Agno SDK and AgentOS runtime baseline: agent/tool execution, delegation, workflow handoffs, approval, durable dispatch and component admission at the frozen revision. |
| pal |
Pal runtime baseline: coordinated knowledge work, direct ingestion, scheduled maintenance, controls and Git synchronization. |
| claude-obsidian |
claude-obsidian runtime: host-directed knowledge workflows, transaction admission, retrieval alternatives and lifecycle context at the frozen commit. |
| SkillRL |
SkillRL runtime baseline: skill-conditioned environment rollouts, policy fitting, failure-triggered bank admission and offline data production. |
| AriGraph |
AriGraph runtime baseline: shipped game decisions, navigation, QA, alternative memory consumers and their control limits at the frozen commit. |
| llm-project-wiki |
LLM Project Wiki runtime baseline: model-mediated bootstrap, wiki-first use, source sync, diff ingest and existing-vault repair. |
| Stash |
Stash runtime baseline: MCP and CLI memory requests, model-driven maintenance and release admission at d34ed430c4348ef741ef95146b488aea8616f4fb. |
| AI-Context-OS |
AI-Context-OS runtime baseline tracing desktop chat, MCP context serving and workspace change admission at the public snapshot. |
| echel |
Echel runtime baseline: canonical initialization, legacy work/context execution, verification and scoped change admission at aee36de4d2d1c404fb36e9ad68cc0a920784ee61. |
| Zikkaron |
Zikkaron runtime baseline: MCP memory operations, direct SQLite host hooks, compaction replay, instruction installation and evaluation at the frozen source revision. |
| SAGE |
SAGE runtime baseline: paired AppWorld rollouts, skill execution, reward-fed policy updates, evaluation and expert-to-SFT admission. |
| quicky-wiki |
Quicky Wiki runtime: source-to-graph compilation, query consumption, client mutation and maintenance at cfeecbffce5b6d0f16533b258a1e88edd86f731e. |
| Apache Maka |
Apache Maka runtime baseline: Host-admitted model/tool turns, boundary controls, delegated and durable work, and plugin capability admission at the frozen commit. |
| Synapptic |
Synapptic runtime baseline: transcript extraction, profile admission, publication, benchmark feedback and optional relay at the frozen revision. |
| Tracecraft |
Tracecraft runtime baseline: bucket-backed coordination, session mirroring and shipped change paths at the frozen SDK revision. |
| arsumbris |
Arsumbris runtime baseline: documented typed-graph, host and agent routes plus the implemented device-config seeder at the frozen entry-repository boundary. |
| llm-context-base |
Runtime baseline of llm-context-base's instruction-driven session, capture, query and maintenance paths at the frozen template revision. |
| Claude Context Guard |
Claude Context Guard's installation, session recovery, checkpoint, lesson, audit and hook routes at the frozen source revision. |
| Synthadoc |
Synthadoc runtime baseline: queued knowledge compilation, bounded answering, maintenance tool loops and alternate admission paths at the frozen Git boundary. |
| TheKnowledge |
TheKnowledge runtime baseline: incremental authorship, durable deposits, retrieval, remote research and instruction admission at the frozen Git boundary. |
| MemoryOS |
MemoryOS runtime baseline across JSON libraries, MCP, Playground, ChromaDB and evaluation at the frozen repository revision. |
| Basic Memory |
Basic Memory runtime baseline: local note acceptance, deferred materialization, context retrieval and Pi checkpoint loops at the frozen revision. |
| beever-atlas |
Beever Atlas runtime: ingestion, wiki maintenance, QA and alternate MCP/control paths at the frozen whole-system boundary. |
| GMemory |
GMemory runtime: shipped benchmark invocations, multi-agent action paths, retained-context delivery and automatic admission at the frozen revision. |
| AI Agents in Depth |
AI Agents in Depth runtime baseline: context-agent invocation, ablation and executor counterpaths, and bounded retry-policy change admission. |
| Agent-R |
Agent-R runtime baseline: environment-driven MCTS, revision-data admission and evaluation at the frozen repository revision. |
| ByteRover CLI |
ByteRover CLI runtime: caller-authored knowledge, daemon admission, deterministic retrieval and alternate mutation paths at the frozen repository boundary. |
| oh-my-pi |
oh-my-pi runtime baseline: CLI session, model/tool control, delegated contexts and alternate effect/admission paths at the frozen commit |
| ATLAS |
ATLAS runtime baseline: shipped chat, extraction, enrichment, MCP and administrative paths at the frozen repository revision. |
| claude-workstream-kit |
Claude Workstream Kit runtime baseline: installed instructions, state reload, model-mediated work, guarded rewrites and change admission. |
| Tardigrade |
Tardigrade runtime baseline tracing atom invocation, governed and code execution, legacy hosting, recovery and actor admission at the frozen commit. |
| llmwiki-marimo |
Runtime baseline of llmwiki-marimo: ingestion, three chat modes, save admission and maintenance at the frozen source revision. |
| MentisDB |
MentisDB runtime baseline: memory admission and read-back, alternate transports, extraction, skill admission and release updates at the frozen Rust/Python boundary. |
| llm-wiki |
LLM Wiki runtime baseline: external-agent maintenance, CLI/MCP retrieval, scaffolding and forcing routes at the frozen commit. |
| CocoIndex |
CocoIndex runtime baseline: incremental indexing, live maintenance, preview, model extraction and shipped retrieval at the frozen revision. |
| Halo |
Halo runtime baseline: shipped trace diagnosis, delegation, bounded execution, desktop follow-ups and improvement handoff at b7f8509745d67b499b4e80efe20ea37c03426a74. |
| Exocomp |
Exocomp runtime baseline: planner inference, contractor processes, tool effects, operator routes and change admission at the frozen source revision. |
| CrewAI |
CrewAI runtime baseline: Crew and standalone Agent execution, mediated tools, Flow control and instruction admission at the frozen revision. |
| Ecdysis |
Ecdysis runtime baseline: failure review, adapter-mediated harness editing, score admission and retained-harness inference at the frozen commit. |
| Decapod |
Decapod runtime baseline: ephemeral CLI/RPC governance, agent-owned work, context delivery, custody and exact-candidate publication at the frozen repository revision. |
| smriti-mcp |
Smriti MCP runtime baseline: caller-driven durable writes, recall, trace consolidation and maintenance at the frozen source revision. |
| Engraph |
Engraph runtime baseline: shipped retrieval, vault admission, maintenance and server controls at f9a95bc96accc792c02ee384d9e6bf768a88c8c8. |
| beads_rust |
beads_rust runtime baseline: CLI issue lifecycle, MCP requests, JSONL admission, instruction installation and executable replacement at the frozen commit. |
| cass_memory_system |
cass_memory_system runtime baseline: context retrieval, reflection admission, caller-assisted onboarding, HTTP tools and host safety hooks at the frozen revision |
| eigenius |
Eigenius runtime baseline: asynchronous prose formalization, separate knowledge admission, and effectful program execution at the frozen revision. |
| LLM-WIKI-MCP |
LLM-WIKI-MCP runtime baseline: shipped ask paths, deterministic retrieval, model synthesis and local change admission at the frozen revision. |
| Compound Engineering Plugin |
Compound Engineering Plugin runtime baseline: host-loaded skills, native implementation, cross-model transactions, shipping and change admission. |
| exo |
Exo runtime baseline tracing managed turns, harness alternatives, capability admission and host self-update at the frozen source revision. |
| Hyperagents |
Hyperagents runtime baseline: code-editing generations, task evaluation, lineage replay and optional selection/ensemble execution at the frozen revision. |
| Closure-SDK |
Closure-SDK current Rust Closure EA runtime: ingestion, corrective feedback, direct memory access, inference and state restoration. |
| pi-posthorse |
Pi Posthorse runtime baseline: foreground reset admission, automatic recovery, request guidance and bounded retained-state retrieval at the frozen extension boundary. |
| incremental-self-improvement |
Incremental self-improvement's documented probabilistic interpreter, self-modification and reward-rate retention loop in the revised FKI-198-94 capture. |
| From Agent Memory to Portable Skills |
Runtime baseline of From Agent Memory to Portable Skills: documented NAMS distillation, gates, review, delivery and bounded repair in the frozen article. |
| GBrain |
GBrain runtime baseline: capture admission and publication, MCP controls, context delivery, delegated tool execution and shared-skill revision at the frozen commit. |
| archie |
Archie runtime-client baseline: persistent model-role selection and zsh launch paths at the frozen revision. |
| JEPA-Anything |
JEPA-Anything runtime baseline: host-model task compilation, deterministic scaffold admission, generated consumers and structural evaluators. |
| nao |
nao runtime baseline: analytics chat, delegated and scheduled execution, tool boundaries and context-change admission at the frozen commit. |
| Sentry Skills |
Sentry Skills runtime baseline: host-consumed instructions, PR repair helpers, approval-scoped triage and skill change admission. |
| Graphiti |
Graphiti runtime baseline tracing episode ingestion, retrieval, shipped consumers, service queues and alternate admission paths at the frozen commit. |
| Phantom |
Phantom runtime baseline: channel and durable chat execution, SDK controls, dynamic capabilities and reflection admission at the frozen revision. |
| harness2 |
Harness² runtime baseline: shipped sweep, recursive harness editing, execution, selection and change admission at the frozen revision. |
| ReasoningBank |
ReasoningBank runtime baseline: shipped web and coding task loops, automatic memory admission and forcing routes. |
| link |
Link runtime baseline: local context retrieval, admission, session delivery and maintenance at cc92f3632121b04063b24eaaf3a8c5429eaf2b57. |
| autocontext |
Autocontext runtime baseline: scenario generations, task revision, context admission, host execution and model/code improvement at the frozen repository revision. |
| PrimeScientist |
PrimeScientist runtime baseline: filesystem-backed research search, benchmark execution, scoring and plan admission at the frozen revision. |
| MemWiki |
MemWiki runtime baseline: scaffold and update CLI, host-agent protocol, maintenance requests and their enforcement limits. |
| xMemory |
xMemory runtime baseline: shipped construction, hierarchy maintenance, retrieval, answer consumption and evaluation at the frozen Git revision. |
| CORAL |
CORAL runtime baseline: coding-agent launch, asynchronous evaluation, feedback, steering and island migration at the frozen commit. |
| academic-research-skills |
Runtime baseline of academic-research-skills: host-mediated research stages, screening fan-out, revision admission and bounded verification controls. |
| ModularRSI |
ModularRSI runtime baseline: module-library evolution, gated generation admission and later solver consumption at the frozen release. |
| AI Research OS Workshop |
AI Research OS Workshop runtime baseline: host-mediated research routing, wiki admission, helper enforcement and downstream consumption. |
| o-o |
o-o runtime baseline: local polyglot document updates, external Claude dispatch, library drivers and change admission at the frozen commit. |
| Instinctual Memory |
Instinctual Memory runtime baseline: journal consolidation, validated changes, retrieval, host context delivery and alternate maintenance paths. |
| jaz-evals |
jaz-evals runtime baseline: configured evaluation, agent adapters, improvement admission and control limits at the frozen parent revision. |
| DocMason |
DocMason runtime baseline: host-owned reasoning, canonical ask closure, staged publication and local change admission. |
| ReframeWeb |
ReframeWeb runtime baseline tracing bound user input, BAML and Codex execution, task-scoped effects and saved Store admission at the frozen commit. |
| Atomic |
Atomic runtime baseline tracing chat, model-authored writes, wiki admission, reports and maintenance at the frozen repository revision. |
| llm-wiki |
LLM Wiki runtime baseline: ingestion, retained-content consumption, alternate drivers and incomplete prompt-change admission at the frozen Git boundary. |
| kgai |
kgai runtime baseline: foreground decision capture, host nudges, optional model fork, configuration approval, synchronization and editor read-back. |
| MerchantBench |
MerchantBench runtime baseline tracing the shipped ReAct/store loop, HTTP controls, alternate clients, hosted evaluation and Hermes launch boundary. |
| Synto |
Synto runtime baseline tracing vault production, publication, query, MCP and exported agent instructions at the frozen commit. |
| SAGE |
SAGE runtime baseline: signed memory admission, retrieval, host-context hooks and operator-controlled change at e2e0242a421afbd356ed135d7d754a35b0ad5f14. |
| SwarmWorld |
SwarmWorld runtime baseline: model planning, authoritative world stepping, human overrides, artifact programs, scenario admission and recorded replay. |
| OpenWiki |
OpenWiki runtime baseline: native and host repository generation, personal ingestion, retrieval and change admission at the frozen commit. |
| Eidetic |
Eidetic runtime baseline: lifecycle extraction and injection, recall, admission gates, logical-index worker and product updates. |
| ExpeL |
ExpeL runtime baseline tracing benchmark acting, reflection, insight admission and evaluation at the frozen Git revision. |
| context-constitution |
Context Constitution runtime baseline: declared context selection, memory admission, compaction and delegation without the executing Letta harness. |
| Letta Code |
Letta Code runtime baseline: foreground provider/tool turns, durable listener approvals, delegation and executable/configuration admission at the frozen commit. |
| SkillNote |
SkillNote runtime baseline: skill admission, host delivery, context hooks and feedback at the frozen source revision. |
| dense-mem |
Dense-Mem runtime baseline: authenticated Remember, session intake, bounded recall and automatic and manager-controlled maintenance at bd736450454e237f1ab6d138f4ae9755916d73f6. |
| pi-self-learning |
pi-self-learning runtime baseline: host-triggered reflection, memory admission, maintenance and future-turn injection at the frozen extension revision. |
| Mem0 |
Mem0 runtime baseline: Python memory and answer loops, direct-write and API controls, hosted delegation and OpenClaw context delivery at the frozen revision. |
| Memori |
Memori runtime baseline: Python and TypeScript interception, service-dependent augmentation, and shipped agent-host memory routes. |
| Napkin |
Napkin's CLI/SDK retrieval, caller-directed mutations, shipped agent skills and benchmark consumption at the frozen source revision. |
| KBLaM |
KBLaM runtime baseline: knowledge-conditioned generation, alternate context routes, dataset admission and adapter training at the frozen research implementation. |
| MemPalace |
MemPalace runtime baseline: live capture, later recall, MCP control boundaries, setup admission and delegated task delivery at the frozen commit. |
| Voyager |
Voyager runtime baseline tracing Minecraft learning, inference, manual decisions and direct JavaScript execution at the frozen commit. |
| Meta-n |
Meta-n runtime baseline: evaluated recursive injections, native and external solving, admission gates and bounded controls. |
| EchoWiki |
EchoWiki runtime baseline tracing Obsidian and CLI ingestion, model-directed wiki writes and editable compilation instructions. |
| CQ |
CQ runtime baseline: host-led knowledge use, MCP and CLI routes, local fallback, remote review, and optional embedding retrieval. |
| binder |
Binder runtime baseline: graph transactions, alternate admission paths and shipped agent callers at e4066d750ac5c6918706f44b12daa261b261106e. |
| Agent Skills for Context Engineering |
Source-grounded runtime baseline of Agent Skills for Context Engineering: skill delivery, research admission, router evaluation and selected persistent-context examples. |
| auto-harness |
auto-harness runtime baseline: coding-agent improvement loop, three benchmark routes and conditional admission at the frozen Git revision. |
| nuggets |
Nuggets runtime baseline: messaging-to-Pi invocation, memory hooks, proactive scheduling and admission limits at the frozen TypeScript revision. |
| wuphf |
Wuphf runtime baseline: broker-driven office turns, scheduled tools, model authoring and route-local approval boundaries. |
| QwenPaw |
QwenPaw runtime baseline: native request orchestration, tool governance, delegated work, alternative harnesses and capability admission at the frozen commit. |
| secure-llm-wiki |
Secure LLM-Wiki runtime: CLI ingestion, model review, write admission, retained-context queries and operator maintenance at dfbc60e37487b61e9a5cbce1271a13576a776d51. |
| Reflexion |
Reflexion's task-specific attempt, evaluation and reflection workflows, including alternate strategies and incomplete dependency paths. |
| IsaacCLupus_mnemosyn_spec |
Runtime baseline of IsaacCLupus_mnemosyn_spec: proposed unified lifecycle and statically inspected legacy bridge at 0fda14eb33d2fc01a4e3969a9ab487b22c4f9d92. |
| Hermes Agent |
Hermes Agent runtime baseline: one-shot invocation, tool-effect controls, delegation, executable extensions and Codex alternate runtime at the frozen commit. |
| llm-wiki |
llm-wiki runtime baseline: host-operated compilation, bounded retrieval, delegation, adapter execution and change admission at the frozen revision. |
| SoL-Pi |
SoL-Pi runtime baseline: configured extension hooks, fused mutations, log reduction, observation replay, compaction and shipped delegation/visual paths. |
| semantic-engine |
Semantic Engine runtime baseline: operator-driven ingestion, embedding, retrieval and alternate display/export paths at the frozen commit. |
| OpenSage |
OpenSage's manager-driven execution, delegation, capability admission and alternate runner paths at the frozen repository revision. |
| OS-Copilot |
OS-Copilot runtime baseline tracing FRIDAY task execution, repair, capability admission and shipped alternate loops at the frozen revision. |
| HippoRAG |
HippoRAG runtime baseline: retained document indexing, graph/dense retrieval, QA, iterative retrieval and MCP entry paths at the frozen revision. |
| SkillWeaver |
SkillWeaver runtime baseline: website exploration, Python skill admission, native execution and alternate consumers at the frozen Git revision. |
| Enoch |
Enoch runtime baseline: chat actions, durable task execution, governed evolution, adapters and update admission at the frozen application boundary. |
| llm-wiki |
llm-wiki runtime baseline: session conversion, backend synthesis, host handoff, retrieval and publication at the frozen commit |
| prime-agent |
Prime Agent's Rust model/tool loop, Python execution boundary, delegation and change-admission routes at the frozen commit. |
| Prove2Me_workspace |
Prove2Me_workspace runtime baseline: instructed solver, proposal audit and project-upload routes, with executable extraction helpers and external verification limits. |
| echoes-vault-opencode |
EchoesVault for OpenCode runtime baseline: restoration, shared writer admission, lifecycle grants and project-runtime installation at the frozen commit. |
| okf-harness |
OKF Harness runtime baseline tracing agent-owned wiki work, deterministic retrieval, alternate control paths and change admission at the frozen commit. |
| Hyalo |
Hyalo runtime baseline: local CLI work, Pi integration, tidy admission and optional Jev suggestions at the frozen revision. |
| Tolaria |
Tolaria runtime baseline: desktop CLI invocation, direct-provider note creation, MCP writes and guidance admission at the frozen revision. |
| Thalo |
Thalo runtime baseline: deterministic context selection, shipped Pi synthesis writeback, Git publication and alternate authoring paths. |
| talorys |
Talorys runtime baseline tracing owner chat, guarded tools, scheduled work and owner/deployment change paths |
| EvoOntology |
EvoOntology runtime baseline: plugin-driven construction, semantic task execution, evaluation and version admission at c94389b764394379086ff25e50d0e0f4396167b4. |
| REM |
REM runtime baseline tracing API admission, context retrieval, worker scheduling and consolidation gaps at the frozen revision. |
| equipa |
EQUIPA runtime baseline: task execution, provider alternatives, gated product admission and prompt/configuration change paths. |
| cludebot |
Cludebot runtime baseline: authenticated chat, memory entry modes, maintenance and dashboard execution at the frozen repository revision. |
| Tendril |
Tendril runtime baseline: desktop-to-agent invocation, capability admission, Deno execution and alternate workspace controls. |
| interview-doc-agent |
Interview Doc Agent's host-interpreted initialization, dossier maintenance and document-generation routes at its frozen prompt boundary. |
| sig-releases |
Sig runtime baseline: declared capture, AI-provider, sharing, skills and telemetry paths in its public release documentation. |
| Signet |
Signet runtime baseline: Pi context delivery and capture, bounded Dreaming maintenance, and alternate admission and executor paths. |
| Agent S |
Agent S runtime baseline: S3 desktop and coding execution, historical hierarchy, experience admission and offline rollout selection. |
| gnosis |
Gnosis runtime baseline: repository-local CLI, doctrine delivery, retrieval, admission and maintenance at the frozen commit. |
| AutoSci |
AutoSci runtime baseline: prompted research, wiki producers and consumers, browser and scheduled writers, and optional Hiera execution at the frozen commit. |
| Spacebot |
Spacebot runtime baseline: conversational execution, delegated workers, autonomous epochs, context control and change admission at the frozen revision. |
| WikiSkill |
WikiSkill runtime baseline: host-driven work requests, native handoffs, feedback admission and two research execution paths. |
| meta-harness |
Meta-Harness runtime baseline: reference harness search, task execution and the experimental Harbor evaluation loop. |
| Company Brain |
Company Brain's Slack turn loop, alternate execution paths, local controls and change admission at the frozen worker revision. |
| supermemory |
Supermemory runtime baseline: public middleware, tool loops and MCP admission paths, with the memory engine opaque. |
| Dynamic Cheatsheet |
Dynamic Cheatsheet runtime baseline tracing generation, curation, retrieval, execution and benchmark continuation at the frozen source revision. |
| Swamp |
Swamp runtime baseline: external-agent launch, typed behavior and workflow execution, approval, extension admission and repository review at the frozen commit. |
| WeKnora |
WeKnora runtime baseline tracing agent turns, RAG, tool controls and skill-image admission at the frozen Git revision. |
| Pond |
Pond runtime baseline: capture, model-directed recall, SQL escape paths and session restoration at the frozen revision. |
| Semiont |
Semiont runtime baseline tracing gathered-context generation, queued annotation work, direct admission and compiler repair at the frozen repository revision. |
| ai-modules |
ai-modules runtime baseline: model-interpreted task and wiki workflows, helper controls, charter interception, deployment and evaluator launch at the frozen commit. |
| eal-bench |
EAL-Bench runtime baseline tracing writer admission, executor scoring, replay and closed-loop write-back at the frozen revision. |
| Cognee |
Cognee runtime baseline: ingestion, recall, session maintenance, decorated consumers and agentic tools at the frozen revision. |
| Fractal |
Fractal runtime baseline: CLI admission, direct workspace delegation, durable turns, interruption and configuration at the frozen revision. |
| playground |
Playground runtime baseline: tenant sandbox execution, bounded jobs, file exchange and the optional native-worker gateway. |
| Kompl |
Kompl runtime baseline: compile orchestration, knowledge admission, query feedback and alternate maintenance paths at the frozen commit. |
| funes |
Funes runtime baseline: external Librarian invocation, retained-library workflow, extraction helpers and mechanical CI at the frozen revision. |
| Continuity |
Continuity runtime baseline: desktop conversation, isolated research, memory admission, plugin capability controls and Threads publication at the frozen source revision. |
| PageIndex |
PageIndex runtime baseline: local indexing and document QA, alternate builders, framework tools and opaque cloud routes. |
| Hindsight |
Hindsight runtime baseline: HTTP reflection, bounded retrieval and synthesis, asynchronous retention, Claude Code hooks and configuration admission at the frozen commit. |
| VLM-wiki |
VLM-wiki runtime baseline: host-mediated compilation and query, direct image calls, deterministic reviews, import and delivery. |
| browzy.ai |
browzy.ai runtime baseline: terminal and library answering, source compilation and retained-change admission at the frozen Git boundary. |
| Siftly |
Siftly runtime baseline: import, model enrichment, category admission and retained-content consumption at the frozen Git revision. |
| Virtual Context |
Virtual Context runtime baseline tracing shipped chat, proxy, paging, import, administration and engagement paths at the frozen repository revision. |
| Voiden |
Voiden runtime baseline: agent-facing MCP request execution, plugin admission and request-backed publishing at the frozen subsystem boundary. |
| OpenViking |
OpenViking runtime baseline: scoped retrieval, context assembly, durable session processing and conditional skill/config admission at the frozen revision. |
| deja-vu |
deja-vu runtime baseline: local history ingestion, host delivery, explicit admission and recovery at the frozen Git boundary. |
| LLM-Wiki-v3 |
LLM-Wiki-v3 runtime baseline: extraction, curated admission, index-backed retrieval and evaluation at the frozen source revision. |
| SkillLift |
SkillLift runtime baseline: task-portfolio evolution, verifier calibration, durable recovery and distinct category/domain drivers at the frozen main-repository revision. |
| Auto-claude-code-research-in-sleep |
Runtime baseline of Auto-claude-code-research-in-sleep: host-driven research stages, reviewer transports, acceptance/resume controls and harness-patch admission at the frozen partial-loop boundary. |
| memex |
Memex runtime baseline tracing Unix-socket admission, per-wiki jobs, delegated Claude work and control limits at the frozen commit. |
| Self-Training-LLM |
Self-Training-LLM runtime baseline: operator-launched generation, SFT/DPO admission, checkpoint consumption and model judging at the frozen source revision. |
| claude-code-dynamic-workflows |
Claude Code dynamic workflows runtime baseline: documented script orchestration, launch controls, worker effects and session-scoped replay. |
| A-MEM |
A-MEM runtime baseline tracing note admission, model-mediated evolution, retrieval and direct mutation at its frozen library boundary. |
| sage-wiki |
Sage-wiki runtime baseline: capture, tiered compilation, retrieval and answer admission, with direct-write, provider and parser alternatives at the frozen commit. |
| openclerk |
OpenClerk runtime baseline tracing agent context, session planning, document admission, semantic modules and evaluation callers at the frozen commit. |
| Sparks |
Sparks runtime baseline: external-agent semantic work, CLI/MCP maintenance, persistence and later consumption at the frozen revision. |
| ai-memex-cli |
Runtime baseline of ai-memex-cli: delegated knowledge maintenance, deterministic feedback, and later-session context delivery at the frozen revision |
| SkillX |
SkillX runtime baseline: trajectory-to-library construction, prompt preparation, exploration and change admission at the frozen commit. |
| lhtb |
LHTB runtime baseline: terminal task execution, continuation, hidden verification and alternate Harbor integration paths at the frozen revision. |
| Agent Workflow Memory |
Agent Workflow Memory runtime baseline: benchmark callers, model action paths and workflow admission at the frozen source revision. |