{"passport":{"unfragile":{"@version":"1.0","version":"2026-05","artifact":{"id":"smithery_rfi-irfos-ternlang","slug":"rfi-irfos-ternlang","name":"Ternary Intelligence Stack","type":"mcp","url":"https://ternlang.com","page_url":"https://unfragile.ai/rfi-irfos-ternlang","categories":["mcp-servers"],"tags":["mcp","model-context-protocol","smithery:rfi-irfos/ternlang"],"pricing":{"model":"open_source","free":true,"starting_price":null},"status":"active","verified":false},"capabilities":[{"id":"smithery_rfi-irfos-ternlang__cap_0","uri":"capability://planning.reasoning.ternary.state.routing.with.trit.decide","name":"ternary-state routing with trit_decide","description":"Implements a three-state decision system (affirm/hold/reject) as a first-class routing primitive instead of binary yes/no forcing. The trit_decide function evaluates agent decisions against evidence sufficiency thresholds, returning a trit (ternary bit) that routes to proceed, wait-for-more-data, or block paths. This prevents false-positive commitments on ambiguous data by introducing a structural 'hold' state that triggers evidence-gathering loops before final decisions.","intents":["I need my agent to pause and gather more evidence when confidence is insufficient instead of guessing","I want to prevent irreversible actions (API calls, database writes) when data is ambiguous","I need to route agent behavior based on confidence levels, not just binary outcomes"],"best_for":["AI agent builders working on high-stakes decisions (financial, medical, compliance)","Teams implementing EU AI Act Article 13/14/15 compliance for high-risk systems","Developers building multi-turn agents that need explicit evidence-gathering loops"],"limitations":["Requires explicit confidence thresholds to be defined per decision type — no automatic tuning","Hold state adds latency (evidence gathering loops extend execution time)","Integration with existing binary-logic systems requires refactoring decision trees"],"requires":["Ternlang runtime v1.0.0+","MCP server integration (Model Context Protocol)","Evidence scoring function or confidence metric from upstream LLM"],"input_types":["structured decision context (JSON)","confidence scores (float 0-1)","evidence payloads (text, structured data)"],"output_types":["trit value (affirm/hold/reject)","routing instruction (proceed/wait/block)","evidence gap report (what data is missing)"],"categories":["planning-reasoning","safety-moderation"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"smithery_rfi-irfos-ternlang__cap_1","uri":"capability://planning.reasoning.consensus.based.multi.agent.trit.consensus","name":"consensus-based multi-agent trit_consensus","description":"Aggregates ternary decisions from multiple agent instances or reasoning paths using a consensus mechanism that produces a single trit output. Rather than averaging confidence scores, trit_consensus evaluates agreement patterns across agents (all affirm, mixed affirm/hold, all reject, etc.) and applies voting rules to produce a robust collective decision. This enables distributed agent architectures where disagreement triggers hold states for human review or additional reasoning.","intents":["I want multiple agents to vote on a decision and only proceed if consensus is strong","I need to detect when agents disagree and escalate to human review","I want to combine diverse reasoning paths (different prompts, models) into a single decision"],"best_for":["Multi-agent systems requiring consensus before high-risk actions","Compliance-heavy workflows needing audit trails of disagreement","Teams using ensemble reasoning (multiple LLM calls with different strategies)"],"limitations":["Consensus computation adds latency proportional to number of agents (N agents = N LLM calls)","Voting rules must be pre-configured; no adaptive consensus weighting based on agent reliability","Requires all agents to return trits in compatible format — heterogeneous agent types need adapters"],"requires":["Ternlang v1.0.0+","Multiple agent instances or reasoning paths producing trit outputs","Voting rule configuration (simple majority, unanimous affirm, etc.)"],"input_types":["array of trit values from multiple agents","confidence scores per agent (optional)","voting rule specification (JSON)"],"output_types":["consensus trit (affirm/hold/reject)","agreement matrix (which agents agreed/disagreed)","dissent report (agents voting differently)"],"categories":["planning-reasoning","safety-moderation"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"smithery_rfi-irfos-ternlang__cap_2","uri":"capability://memory.knowledge.ternary.vector.embeddings.with.trit.vector","name":"ternary vector embeddings with trit_vector","description":"Generates and operates on ternary vector representations where each dimension encodes a trit (affirm/hold/reject) instead of continuous float values. This enables semantic search and similarity operations that respect three-state logic: two vectors are similar if they agree on affirm/hold/reject across dimensions, with hold dimensions treated as 'don't care' wildcards. Useful for retrieving similar past decisions or evidence patterns from a ternary decision history.","intents":["I want to find similar past decisions in my agent's history using ternary logic","I need to cluster decisions by agreement patterns (which dimensions agents affirmed vs held)","I want semantic search over decision histories that respects confidence levels"],"best_for":["Agents with persistent decision logs needing similarity-based retrieval","Compliance systems auditing decision patterns across time","Multi-agent systems analyzing agreement/disagreement clusters"],"limitations":["Ternary vectors are sparser than float embeddings — may require larger vector dimensions for equivalent expressiveness","Similarity metrics for ternary vectors are not standardized; custom distance functions needed","Integration with standard vector databases (Pinecone, Weaviate) requires adapter layers"],"requires":["Ternlang v1.0.0+","Decision history with trit outputs","Vector dimension specification (typically 256-1024 dimensions)"],"input_types":["sequence of trit values (affirm/hold/reject)","decision context (metadata, timestamps)","query trit vector"],"output_types":["ternary vector (trit per dimension)","similarity scores (0-1, respecting hold semantics)","ranked list of similar past decisions"],"categories":["memory-knowledge","data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"smithery_rfi-irfos-ternlang__cap_3","uri":"capability://planning.reasoning.mixture.of.experts.orchestration.with.moe.orchestrate","name":"mixture-of-experts orchestration with moe_orchestrate","description":"Routes agent tasks to specialized expert sub-agents based on decision type and evidence patterns using a gating network that outputs ternary routing decisions. The MoE-13 deliberation engine maintains 13 expert agents (financial, medical, legal, technical, etc.) and uses trit outputs to route: affirm routes to execution, hold routes to multi-expert consensus, reject routes to escalation. Gating decisions are themselves ternary, enabling hold states when task classification is ambiguous.","intents":["I want different expert agents to handle different decision types (financial vs medical vs legal)","I need to route ambiguous tasks to multiple experts for consensus before proceeding","I want to escalate decisions that no expert can confidently handle"],"best_for":["Multi-domain agent systems (e.g., compliance platforms handling finance + legal + technical decisions)","Organizations with specialized teams that should be consulted for different decision types","High-stakes systems where task misclassification is costly"],"limitations":["MoE-13 is fixed architecture — adding custom experts requires framework modification","Routing latency scales with number of experts consulted (hold states may trigger all 13 experts)","Expert specialization is pre-trained; no online learning to adapt expert boundaries"],"requires":["Ternlang v1.0.0+ with MoE-13 module","Task classification input (domain tags or natural language description)","Expert agent definitions (prompts, capabilities, confidence thresholds)"],"input_types":["task description (text)","domain classification (enum or string)","evidence payload (structured data)"],"output_types":["expert routing decision (which expert(s) to consult)","consensus trit from selected experts","expert confidence scores per domain"],"categories":["planning-reasoning","tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"smithery_rfi-irfos-ternlang__cap_4","uri":"capability://automation.workflow.ternary.aware.agent.execution.with.ternlang.run","name":"ternary-aware agent execution with ternlang_run","description":"Executes agent loops with native support for ternary control flow: affirm proceeds to next step, hold triggers evidence-gathering sub-loops (additional tool calls, web searches, expert consultation), reject terminates and escalates. The runtime maintains a ternary state machine where transitions are guarded by trit outputs from decision points. Integrates with MCP servers for tool access and maintains execution traces for audit compliance.","intents":["I want my agent to automatically gather more evidence when uncertain instead of guessing","I need execution traces showing why the agent held vs affirmed vs rejected at each step","I want to build agents that respect EU AI Act requirements for explainability and human oversight"],"best_for":["Compliance-heavy agent systems (financial, medical, legal)","Teams building auditable AI workflows with clear decision trails","Developers migrating from binary ReAct agents to ternary reasoning"],"limitations":["Hold loops add unpredictable latency — execution time depends on evidence-gathering depth","Requires explicit evidence-gathering strategies; no automatic evidence source discovery","State machine complexity grows with number of decision points (N decisions = 3^N possible paths)"],"requires":["Ternlang v1.0.0+","MCP server integration for tool access","Agent definition with ternary decision points (trit_decide calls)","Evidence-gathering tool definitions (web search, database queries, expert APIs)"],"input_types":["agent definition (JSON or Ternlang DSL)","initial task/query (text)","tool specifications (MCP server endpoints)"],"output_types":["final decision (affirm/hold/reject)","execution trace (JSON with all decision points and trits)","evidence chain (what data was gathered at each hold state)"],"categories":["automation-workflow","planning-reasoning"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"smithery_rfi-irfos-ternlang__cap_5","uri":"capability://safety.moderation.eu.ai.act.compliance.auditing.with.trit.audit","name":"eu ai act compliance auditing with trit_audit","description":"Generates compliance reports for Articles 13, 14, and 15 of the EU AI Act by analyzing agent execution traces and ternary decision logs. Checks for: (13) transparency of high-risk AI system decisions, (14) human oversight mechanisms (hold states triggering escalation), (15) accuracy and robustness (consensus mechanisms, evidence thresholds). Produces structured audit reports mapping each agent decision to compliance requirements and evidence chains.","intents":["I need to prove my AI system complies with EU AI Act Articles 13/14/15","I want to generate audit reports showing human oversight of high-risk decisions","I need to document decision rationales and evidence for regulatory review"],"best_for":["EU-based organizations deploying high-risk AI systems","Compliance teams needing automated audit trail generation","Legal/regulatory teams reviewing AI system governance"],"limitations":["Audit scope limited to decisions made within Ternlang runtime — external agent decisions not captured","Compliance interpretation is framework-specific; legal review still required for final certification","Requires detailed execution traces; sparse logging reduces audit quality"],"requires":["Ternlang v1.0.0+","Complete execution traces from ternlang_run (with all decision points)","Agent definitions documenting human oversight mechanisms","EU AI Act article mapping configuration"],"input_types":["execution trace (JSON from ternlang_run)","agent definition with oversight annotations","decision log (trit outputs + evidence)"],"output_types":["compliance report (JSON/PDF)","article-by-article mapping (which decisions satisfy Art. 13/14/15)","evidence chain per decision (for regulatory review)"],"categories":["safety-moderation","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"smithery_rfi-irfos-ternlang__cap_6","uri":"capability://code.generation.editing.bet.vm.compilation.and.execution","name":"bet vm compilation and execution","description":"Compiles Ternlang agent definitions to BET (Binary Execution Ternary) bytecode using a real compiler (not an interpreter or simulator), enabling deterministic execution and formal verification. The VM executes ternary state machines with guaranteed semantics: affirm/hold/reject transitions are atomic, no race conditions in multi-agent consensus, and execution traces are cryptographically hashable for audit immutability. Supports both local execution and distributed deployment across multiple nodes.","intents":["I want deterministic agent execution with formal guarantees, not probabilistic behavior","I need to verify agent behavior before deployment (formal verification of ternary logic)","I want immutable audit trails using cryptographic hashing of execution traces"],"best_for":["High-assurance systems (financial, medical, critical infrastructure)","Teams needing formal verification of agent behavior","Organizations requiring cryptographically-signed audit trails"],"limitations":["BET VM is Ternlang-specific — cannot execute arbitrary Python/JavaScript code","Compilation adds build-time overhead; agent definitions must be pre-compiled","Formal verification is limited to ternary logic; LLM outputs are still non-deterministic"],"requires":["Ternlang v1.0.0+ compiler","Agent definition in Ternlang DSL (not arbitrary code)","BET VM runtime (local or distributed)"],"input_types":["Ternlang agent definition (DSL syntax)","Compilation configuration (optimization level, verification options)"],"output_types":["BET bytecode (binary)","Execution trace (deterministic, hashable)","Formal verification report (if enabled)"],"categories":["code-generation-editing","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"smithery_rfi-irfos-ternlang__cap_7","uri":"capability://tool.use.integration.30.free.tools.via.mcp.integration","name":"30 free tools via mcp integration","description":"Provides 30 pre-built tools (web search, database queries, API calls, document parsing, etc.) accessible via Model Context Protocol (MCP) servers without API keys or authentication. Tools are exposed as MCP resources that agents can discover and invoke dynamically. Each tool returns structured data compatible with ternary decision logic (confidence scores, evidence payloads). Tools are stateless and can be deployed locally or accessed via public MCP servers.","intents":["I want to build agents without managing API keys or authentication","I need a standard set of tools for evidence gathering (search, lookup, parsing)","I want tools that return structured data compatible with ternary reasoning"],"best_for":["Developers prototyping agents quickly without infrastructure setup","Teams building compliance systems needing deterministic tool behavior","Open-source projects requiring no-cost tool access"],"limitations":["30 tools are fixed set — no custom tool definitions in v1.0.0","Tool availability depends on MCP server uptime (no SLA guarantees for free tier)","Some tools have rate limits or data freshness constraints (web search results may be cached)"],"requires":["Ternlang v1.0.0+","MCP server access (local or remote)","No API keys required"],"input_types":["tool name (string)","tool parameters (JSON, tool-specific schema)"],"output_types":["structured tool result (JSON)","confidence score (float 0-1)","evidence payload (text, structured data)"],"categories":["tool-use-integration","search-retrieval"],"confidence":0.5,"matches":0,"success_rate":0}],"trust":{"score":49,"verified":false,"data_access_risk":"high","permissions":["Ternlang runtime v1.0.0+","MCP server integration (Model Context Protocol)","Evidence scoring function or confidence metric from upstream LLM","Ternlang v1.0.0+","Multiple agent instances or reasoning paths producing trit outputs","Voting rule configuration (simple majority, unanimous affirm, etc.)","Decision history with trit outputs","Vector dimension specification (typically 256-1024 dimensions)","Ternlang v1.0.0+ with MoE-13 module","Task classification input (domain tags or natural language description)"],"failure_modes":["Requires explicit confidence thresholds to be defined per decision type — no automatic tuning","Hold state adds latency (evidence gathering loops extend execution time)","Integration with existing binary-logic systems requires refactoring decision trees","Consensus computation adds latency proportional to number of agents (N agents = N LLM calls)","Voting rules must be pre-configured; no adaptive consensus weighting based on agent reliability","Requires all agents to return trits in compatible format — heterogeneous agent types need adapters","Ternary vectors are sparser than float embeddings — may require larger vector dimensions for equivalent expressiveness","Similarity metrics for ternary vectors are not standardized; custom distance functions needed","Integration with standard vector databases (Pinecone, Weaviate) requires adapter layers","MoE-13 is fixed architecture — adding custom experts requires framework modification","builder identity is not verified yet","no observed match outcomes yet"],"rank_breakdown":{"adoption":0.7568475961712423,"quality":0.51,"ecosystem":0.38999999999999996,"match_graph":0.25,"freshness":0.5,"weights":{"adoption":0.25,"quality":0.25,"ecosystem":0.15,"match_graph":0.23,"freshness":0.12}},"observed_outcomes":{"matches":0,"success_rate":0,"avg_confidence":0,"top_intents":[],"last_matched_at":null},"maintenance":{"status":"active","updated_at":"2026-05-24T12:16:28.138Z","last_scraped_at":"2026-05-03T15:18:25.566Z","last_commit":null},"community":{"stars":null,"forks":null,"weekly_downloads":null,"model_downloads":null,"model_likes":null}},"distribution":{"claim_url":"https://unfragile.ai/submit?claim=rfi-irfos-ternlang","compare_url":"https://unfragile.ai/compare?artifact=rfi-irfos-ternlang"}},"signature":"RT3FQvdSVLlxnnkk5mgamIXrl0yyTKnrzXcuIwRlFPCQXpckZ6ESbkqp7p8YTOu3m0vD9d/gCX3ReqDwu0p0Bg==","signedAt":"2026-06-20T18:57:05.453Z","signedBy":"unfragile.ai","version":1},"_links":{"self":"https://unfragile.ai/api/v1/passport/rfi-irfos-ternlang","artifact":"https://unfragile.ai/rfi-irfos-ternlang","verify":"https://unfragile.ai/api/v1/verify?slug=rfi-irfos-ternlang","publicKey":"https://unfragile.ai/api/v1/trust-passport-public-key","spec":"https://unfragile.ai/trust","schema":"https://unfragile.ai/schema.json","docs":"https://unfragile.ai/docs"}}