| 09ed9f03b74c6741… | A CXU identifier is defined as the string prefix "1220" concatenated with the SHA-256 hash of the claim, producing a content-addressable identifier of 68 characters so that identical claims map deterministically to the same ID. | active | contextual | yes | |
| 61e234b493e5ff83… | Every CXU stored in the graph follows a defined structure containing a content-addressable identifier, claim text, supporting quotes, classification fields, assumption metadata, keywords, confidence, and lifecycle state so that each unit is standardized for storage and retrieval. | active | contextual | yes | |
| 9a0ed9a2d95ec08f… | To draft an NDA for a new biotech prospect, the graph traversal should begin with the NDA contract standard, follow customization links from Celldex, and then apply company policy constraints so that the draft includes required approved content. | active | contextual | yes | |
| f266da8553b20a4c… | To determine which platform capabilities the Patrick financial analysis project uses, the graph traversal should start at the Patrick project node and follow IMPLEMENTS relationships to the PYRANA platform core and code interpreter components. | active | contextual | yes | |
| c1e8caa58fb00a93… | To prepare a proposal for a new pharma client, the graph traversal should move from sales value propositions to life sciences clients, then to scoped pharma projects such as TOLMAR and JNJ, and finally to SOW contract standards so that prior vertical work and templates inform the proposal. | active | contextual | yes | |
| 6382ea183f63dbd7… | Cross-set relationships enable multi-hop queries that combine knowledge across graph sets so that users can answer complex business requests by traversing linked entities and relationship types. | active | contextual | yes | |
| 8518846ace6bf1c5… | In the example CXU, the PyranaHarness implements native LLM tool calling via LiteLLM by executing a Turn 0 plan followed by iterative tool-calling loops until the agent finalizes an AgentOutputEnvelope. | active | contextual | yes | |
| 5479648658fa535d… | The example PYRANA_PLATFORM CXU classifies the PyranaHarness statement as axiom knowledge and a specification claim with confidence 0.98. | active | contextual | yes | |
| b125466d20815975… | Confidence scores below 0.40 indicate speculative content that should be used only for gap-filling, including draft ideas and unvalidated hypotheses. | active | contextual | yes | |
| 7a3d41afc5b44d89… | Confidence scores from 0.40 to 0.59 indicate low confidence and should be marked as assumptions, exemplified by the market estimate that the TAM for agentic platforms is $50B by 2028. | active | contextual | yes | |
| ede6440fd25cfb56… | Confidence scores from 0.60 to 0.79 indicate moderate confidence and possible need for updating, exemplified by the estimate that a standard implementation takes 6 to 8 weeks based on limited projects. | active | contextual | yes | |
| 50e47263972fdc30… | Confidence scores from 0.80 to 0.94 indicate high confidence from a reliable source, exemplified by the statement that Patrick's financial analysis covers more than 50 subsidiaries from an SOW. | active | contextual | yes | |
| b62dda545c7d5339… | Confidence scores from 0.95 to 1.0 indicate a verified fact from an authoritative source, such as the statement that PYRANA uses PostgreSQL, Neo4j, and Qdrant from CLAUDE.md. | active | contextual | yes | |
| 71d6e2e3a924385c… | Prescribed knowledge should be used for standards, policies, and approved language, and this category is regulated and requires explicit approval. | active | contextual | yes | |
| 306894339f506158… | Derived knowledge should be used for conclusions drawn from data or experience, and this category follows the standard approval workflow. | active | contextual | yes | |
| a7d03cdc65de35c6… | Axiom knowledge should be used for stable facts that rarely change, and this category follows a standard auto-approve workflow. | active | contextual | yes | |
| 2b284da298cecf27… | The lifecycle of a CXU progresses through DRAFT, ACTIVE, INACTIVE, and VERSIONED states so that the graph can track the status and evolution of each knowledge unit. | active | contextual | yes | |
| 9982f4046f873fcc… | The confidence field in a CXU is a floating-point value ranging from 0.0 to 1.0 so that the certainty of extracted knowledge can be quantified. | active | contextual | yes | |
| 6b463a0301159de0… | The keywords field in a CXU stores a list of terms used for full-text search and filtering so that CXUs can be efficiently retrieved and narrowed by topic. | active | contextual | yes | |
| dd09017bb297c5ec… | The parent_assumption_ids field stores a list of dependent assumption chains so that assumptions can be traced to the assumptions they rely on. | active | contextual | yes | |
| d5a60868937c5500… | The is_assumption field in a CXU is a boolean that must be true for projections and estimates so that uncertain forward-looking content is explicitly marked. | active | contextual | yes | |
| db2bdaa5f09f3bed… | The CXU claim_type field is constrained to definition, requirement, procedure, specification, or relationship values so that the form of each claim is explicitly classified. | active | contextual | yes | |
| a0dafd31b8a22c73… | The CXU knowledge_type field is constrained to axiom, derived, or prescribed values so that stored knowledge is categorized by its epistemic status. | active | contextual | yes | |
| cd4dc32cf8704873… | The supporting_quotes field in a CXU must contain a list of 3 to 6 verbatim source quotes so that each claim is directly evidenced by source text. | active | contextual | yes | |
| eab97c6c0537c057… | The claim field in a CXU must be a standalone assertion with a length of 15 to 70 words so that each unit remains self-contained and atomic. | active | contextual | yes | |
| d94e39ad1a33a164… | A CXU identifier is generated as the prefix 1220 plus the SHA-256 hash of the claim, producing a content-addressable identifier that is 68 characters long. | active | contextual | yes | |
| f4b5545c2b173316… | Every CXU stored in the graph uses a defined structure that includes a content-addressable cxu_id, a standalone claim, supporting quotes, knowledge and claim classifications, assumption tracking, keywords, confidence, and lifecycle state so that each unit is consistently represented. | active | contextual | yes | |
| b2c04489fdd1a270… | A cross-set query can support drafting an NDA for a new biotech prospect by starting from the NDA template, referencing past biotech NDA customizations for Celldex, and applying company policy constraints to determine required content. | active | contextual | yes | |
| 95336c8ec6a68847… | A cross-set query can determine which platform capabilities the Patrick financial analysis project uses by traversing from the project node to the implemented platform components, specifically PYRANA platform core and code interpreter capabilities. | active | contextual | yes | |
| 0f30753a0802b57c… | Cross-set query patterns enable multi-hop retrieval for preparing a proposal for a new pharma client by traversing value propositions, life sciences clients, scoped pharma projects, and SOW contract standards so that prior vertical knowledge, project examples, and template standards can be combined. | active | contextual | yes | |
| 5315731400f819e4… | Section 5.1 states that every CXU stored in the graph follows a defined structure, indicating that CXU anatomy is standardized within the graph design. | active | contextual | yes | |
| 2fab662861345769… | To draft an NDA for a new biotech prospect, the query should start from CONTRACT_STANDARDS:NDA, traverse through CUSTOMIZED_FROM to CLIENTS:CELLDEX, and then traverse through CONSTRAINS to COMPANY_POLICIES so that required inclusions can be identified from past customizations and policy constraints. | active | contextual | yes | |
| 0432e43da3950d5b… | To determine what platform capabilities the Patrick financial analysis project uses, the query should traverse from PROJECTS:PATRICK_FINANCIAL_ANALYSIS through IMPLEMENTS to PYRANA_PLATFORM:CORE and PYRANA_PLATFORM:CODE_INTERPRETER. | active | contextual | yes | |
| af6f2edef94a5b11… | To prepare a proposal for a new pharma client, the query should traverse from SALES_POSITIONING:VALUE_PROPS through POSITIONS_FOR to CLIENTS:LIFE_SCIENCES, then through SCOPED_BY to past pharma projects, and finally through DELIVERED_UNDER to CONTRACT_STANDARDS:SOW templates. | active | contextual | yes | |
| cc44f6eed2e934b1… | The BASED_ON relationship links PROJECTS:deliverables to PYRANA_PLATFORM with many:many cardinality when deliverables are based on platform components so that deliverables can reference multiple underlying components. | active | contextual | yes | |
| c1b24c1128776742… | The EVOLVED_FROM relationship links SALES_POSITIONING version 2 to SALES_POSITIONING version 1 with 1:1 cardinality when positioning evolves over time so that each newer positioning artifact can reference its direct predecessor. | active | contextual | yes | |
| ee6029aac1714933… | The FUNDS relationship links FINANCIAL_OPS to PROJECTS with 1:many cardinality when financial allocations fund projects so that one financial source can fund multiple projects. | active | contextual | yes | |
| dd997a4d4c4ae767… | The PROTECTS relationship links IP_PATENTS to PYRANA_PLATFORM with many:many cardinality when patents protect platform innovations so that intellectual property can be associated with multiple innovations. | active | contextual | yes | |
| 7f74c7f4ebce25ed… | The DEMONSTRATES relationship links MARKETING_CONTENT to PYRANA_PLATFORM with many:many cardinality when content demonstrates platform capabilities so that marketing assets can showcase multiple capabilities and vice versa. | active | contextual | yes | |
| 5631e0134118546b… | The POSITIONS_FOR relationship links SALES_POSITIONING to CLIENTS with many:many cardinality when positioning materials target client segments so that multiple positioning assets can target multiple client groups. | active | contextual | yes | |
| f1e1d5b591b8ae4b… | The CONSTRAINS relationship also links SECURITY_COMPLIANCE to PROJECTS with 1:many cardinality when security requirements constrain delivery so that one security requirement can constrain multiple project deliveries. | active | contextual | yes | |
| d251a60bedafbc51… | The CONSTRAINS relationship links COMPANY_POLICIES to PROJECTS with 1:many cardinality when policies constrain project execution so that one policy can constrain multiple projects. | active | contextual | yes | |
| 8d86f6de54eb2c9f… | The IMPLEMENTS relationship links PROJECTS to PYRANA_PLATFORM with many:many cardinality when projects implement platform capabilities so that multiple projects and platform capabilities can be connected bidirectionally. | active | contextual | yes | |
| 860ee61b9cb68481… | The DELIVERED_UNDER relationship links PROJECTS to CONTRACT_STANDARDS:SOW with 1:1 cardinality when a project is delivered under a specific statement of work so that each project maps to one specific SOW. | active | contextual | yes | |
| b3e2b561302000a3… | The SCOPED_BY relationship links PROJECTS to CLIENTS with many:1 cardinality when a project belongs to a client so that multiple projects can be associated with one client. | active | contextual | yes | |
| 9d6d60ecb6bb82cd… | The GOVERNS relationship links CONTRACT_STANDARDS to CLIENTS with 1:many cardinality when an MSA or TOS governs a client relationship so that one contract standard can govern multiple client relationships. | active | contextual | yes | |
| 739e97f2ab8173ec… | The expected extraction volume for the TEAM_NETWORK set is approximately 50 to 100 CXUs so that planning for this set assumes a relatively large number of team-network knowledge units. | active | contextual | yes | |
| 87abefd944544e2f… | TEAM_NETWORK extraction should use Eric LinkedIn Connections.csv, team-related shared-drive documents, and Granola meeting notes as source materials so that team and network relationships are grounded in those records. | active | contextual | yes | |
| b34e8b9664afde9f… | Knowledge extracted for the TEAM_NETWORK set should be classified as axiomatic knowledge so that the set uses the designated knowledge type for team and relationship facts. | active | contextual | yes | |
| 67b0c0b1f84b8166… | CXUs extracted for the TEAM_NETWORK set should use the claim types definition and relationship so that team-network knowledge is represented in the prescribed forms for this set. | active | contextual | yes | |