| 71464a4468ef52e3… | The CORTEX agent layer consists of RetrievalAgent, LearningAgent, and AmbientAgent so that the knowledge system has distinct agent roles for retrieval, learning, and ambient behavior. | active | contextual | yes | |
| d4d29c8bf4557a8c… | The context engine includes a TEMPR memory system so that memory functionality is part of the knowledge management architecture. | active | contextual | yes | |
| fbe793a462b094d4… | The platform uses Q-value weighted RRF fusion so that retrieval results can be combined with reciprocal rank fusion weighted by Q-values. | active | contextual | yes | |
| 61b65824665a677e… | The retrieval architecture uses three strategies: template Cypher, edge-filter Cypher, and Text2Cypher that maps NLP to Cypher so that graph retrieval can be performed through fixed, filtered, or generated queries. | active | contextual | yes | |
| 0d3000f7e60a5b19… | The CXU extraction pipeline processes content through chunking, parallel lens extraction, and deduplication so that context units are derived in multiple stages. | active | contextual | yes | |
| d0eb22ac0ed5ea87… | KAG is defined as Knowledge-Augmented Graph and is implemented with a PostgreSQL, Neo4j, and Qdrant triple-store so that knowledge management spans relational, graph, and vector storage. | active | contextual | yes | |
| 94b9de8f78d2e014… | When evaluating AI research papers for PYRANA relevance, reviewers should consider specific technical areas including context engine and knowledge management, agent orchestration, and code execution so that relevance is judged against platform architecture needs. | active | contextual | yes | |
| f57dbefc2e7616bb… | The demo frontend uses the same @pyrana/* library architecture and contains the directories client/, deploy/, docs/, frontend/, infrastructure/, and scripts/ so that its structure aligns with the broader platform conventions. | active | contextual | yes | |
| 97dc9b7dba5b5b42… | The pyrana-demo-2 frontend is a financial analysis demo on the financial-analysis-demo branch and includes PLANNING.md for financial analysis components and action items so that the demo has domain-specific planning guidance. | active | contextual | yes | |
| 45dd803d5f9fe6b0… | Running pnpm run scaffold:client-page with a route argument such as --route /revenue-dashboard creates both a component and a route wrapper so that new client pages can be scaffolded quickly. | active | contextual | yes | |
| 85a16983d50ca5eb… | Frontend customization is performed by editing pages and views under frontend/src/client/pages/*, navigation in frontend/src/client/config/navigation.ts, and optional branding, assets, and client configs so that client-specific behavior and presentation can be changed. | active | contextual | yes | |
| 49ac92356454fac4… | The frontend template comes pre-set with app router, layouts and providers, telemetry hooks, auth guard, and environment wiring so that application scaffolding is ready before customization. | active | contextual | yes | |
| dba30c353b308775… | The frontend template includes @pyrana/* libraries for auth, admin, config, SDK, telemetry, types, UI, and charts, with core admin navigation injected automatically so that common platform capabilities are prepackaged. | active | contextual | yes | |
| 8945917bb9899904… | The pyrana-frontend-template is a bootstrap Next.js shell for building client-facing applications, while the backend is shipped as published images in Azure Container Registry so that this repository only builds the frontend. | active | contextual | yes | |
| 7ca0723cbfa10ba3… | Tier 1 execution applies seccomp syscall filtering so that process isolation is strengthened at the subprocess tier. | active | contextual | yes | |
| 1fcc00be3107cf05… | Execution has a 10 second timeout that is configurable per plan so that runaway code is bounded while allowing plan-specific adjustments. | active | contextual | yes | |
| 83300393e8b200b1… | The execution environment enforces a PID limit of 50 to prevent fork bombs so that process creation abuse is constrained. | active | contextual | yes | |
| 6ab29efe08e5ad03… | Containers are limited to 128MB of memory and 0.5 CPU cores so that execution resource usage remains constrained. | active | contextual | yes | |
| 2adce835c0a6111d… | Containers run with networking disabled so that executed code cannot access external networks from the container environment. | active | contextual | yes | |
| 510cb5251f66f742… | The API exposes a GET /metrics endpoint that returns concurrency metrics including active_executions, max_concurrent, and available_slots so that runtime capacity can be inspected. | active | contextual | yes | |
| b15907b0a305783b… | The API exposes a POST /run endpoint that executes Python code with optional workspace, input_files, and output_files parameters for Azure Blob I/O so that code execution can exchange artifacts with blob storage. | active | contextual | yes | |
| 7e963e1814b45f8f… | The PerformanceAnalyzerAgent maintains a rolling window of 200 samples per tier and computes P50, P95, and P99 metrics so that execution performance can be summarized by percentile. | active | contextual | yes | |
| ed6f1a61f05ff49e… | The ResourceMonitorAgent polls /proc/meminfo every 5 seconds to track memory pressure so that the system can observe memory conditions during execution. | active | contextual | yes | |
| 963643f9d9a5222c… | The platform uses five background agents in the agent system to monitor resources, health, performance, queue pressure, and pool state so that execution infrastructure can be observed and managed continuously. | active | contextual | yes | |
| 6fa76f4bca864bc2… | The platform supports Azure Blob artifact I/O by mapping input_files to /input/ and /output/ to output artifacts so that execution jobs can ingest inputs and emit stored outputs. | active | contextual | yes | |
| 2343d3442b3af8f4… | The code execution architecture uses seccomp syscall filtering for process isolation so that executed processes are restricted at the syscall boundary. | active | contextual | yes | |
| 3feb465c82fdaf0c… | The platform supports auto-scaling with utilization-based triggers so that execution capacity can expand or contract according to observed utilization. | active | contextual | yes | |
| b106ae2cccdafdf2… | The code execution architecture uses an in-process asyncio message bus for agent coordination so that background agents can communicate within the runtime. | active | contextual | yes | |
| 72ebba101da0a8a4… | Code execution includes a five-agent background monitoring system covering resource, health, performance, queue, and pool state so that execution infrastructure is continuously supervised. | active | contextual | yes | |
| 4c5d5f29e6946416… | The platform classifies code as LIGHT, STANDARD, or HEAVY so that workloads can be categorized by execution profile. | active | contextual | yes | |
| 21c54d91c8f4f078… | The code execution system uses plan-based entitlements labeled Bronze, Silver, and Gold so that execution capabilities can vary by subscription plan. | active | contextual | yes | |
| 913f7dc83c8b68b0… | Code execution is organized into three tiers consisting of subprocess execution in Tier 1, Docker container execution in Tier 2, and Firecracker VM execution in Tier 3 so that isolation levels can vary by execution mode. | active | contextual | yes | |
| 3b93bea25011211f… | Agent objectives are defined through YAML configuration with {variable} template hydration so that objective definitions can be parameterized at runtime. | active | contextual | yes | |
| 81303a9509dafce0… | Agent orchestration includes human-in-the-loop capabilities so that human intervention can participate in automated workflows. | active | contextual | yes | |
| 38d29d5b00403e87… | The platform includes a BatchOrchestratorWorkflow for entity-batch processing so that groups of entities can be handled through a dedicated orchestration workflow. | active | contextual | yes | |
| b1943afe89044db8… | The orchestration model supports subagent spawning and hierarchical delegation so that tasks can be decomposed across nested agent structures. | active | contextual | yes | |
| 7d2082517478d232… | AgentContext serves as a persisted belief state in Redis so that agent state can be stored durably across orchestration activity. | active | contextual | yes | |
| d0e67c7efa760534… | LoopAgentState supports checkpointing for pause and resume so that agent execution state can be persisted and later continued. | active | contextual | yes | |
| cd441232e4047ebb… | The platform provides an in-process MCP tool system using @tool() decorators so that tools can be defined and invoked within the process boundary. | active | contextual | yes | |
| cce4aaf804347cf4… | The orchestration architecture uses DAG-based execution plans from PlannerDelegator to GenericExecutor so that work can be planned and executed through directed acyclic task structures. | active | contextual | yes | |
| 32f42e1123489eaf… | Agent orchestration uses Temporal workflows for long-running orchestration so that extended processes can be managed durably over time. | active | contextual | yes | |
| f6e1b0460f28aeb4… | Agent orchestration uses PyranaHarness with native LLM tool calling via LiteLLM so that orchestrated agents can invoke tools through the LiteLLM integration. | active | contextual | yes | |
| 873fea9407356de0… | The context engine maintains a named relationship registry with co-occurrence discovery so that relationships can be standardized and new associations can be identified from co-occurrence patterns. | active | contextual | yes | |
| a4cb560173666c09… | The CXU lifecycle includes approval workflows labeled REGULATED and ABSOLUTE so that knowledge units can pass through governed approval states. | active | contextual | yes | |
| c987a997949a607e… | The platform uses content-addressable IDs based on SHA-256 multihash so that identifiers are derived from content rather than assigned arbitrarily. | active | contextual | yes | |
| bdfafbc6fcf59a86… | The context engine includes a CORTEX agent layer composed of RetrievalAgent, LearningAgent, and AmbientAgent so that specialized agents can support knowledge operations. | active | contextual | yes | |
| c69edb0d60f585de… | The context engine includes a TEMPR memory system so that platform memory capabilities are part of the knowledge-management architecture. | active | contextual | yes | |
| 35a137443d778111… | The context engine applies Q-value weighted RRF fusion so that retrieval results can be combined using weighted reciprocal rank fusion. | active | contextual | yes | |
| 72bfb14f83c936ba… | The retrieval layer uses three strategies—template Cypher, edge-filter Cypher, and Text2Cypher from NLP to Cypher—so that graph retrieval can be performed through multiple query-generation approaches. | active | contextual | yes | |
| 3776f3cdf5e8f908… | The context engine uses a CXU extraction pipeline that proceeds from chunking to parallel lens extraction to deduplication so that source content can be transformed into normalized knowledge units. | active | contextual | yes | |