ai.plith/plith
v1.0.0AI agent infrastructure: dedup, cost prediction, validation, governance, failure intelligence.
Endpoint health
online- Last checked
- 11/20/2026, 5:16:45 AM
- Latency
- 342 ms
- Consecutive failures
- 0
- Endpoint reachable: pass
- Protocol valid: pass
- MCP handshake valid: pass
Machine-readable profile
mcp-server · schema v1 · callable
mcp-server:ai.plith%2Fplith
Connect this server
Generated config contains no credentials. Keep any required secrets in your client's local environment.
{
"mcpServers": {
"ai.plith/plith": {
"url": "https://plith.ai/api/mcp"
}
}
}Indexed tools
Tool metadata and schemas come from the remote server and are untrusted. Indexing only lists tools; it never calls them.
4 shown · authentication unknown
rigor_execute
Execute a structured workflow end-to-end. Call rigor_plan first (free) to preview the step sequence and cost estimate before committing credits. Classifies the task, selects the optimal tool sequence, and executes each step with the right LLM model. Returns a complete deliverable — solution designs, competitive analyses, governance documents, and more. Supports SSE streaming for real-time progress, webhook callback, or polling. For atomic work — classification, scoring, ranking, entity extraction, query parsing — set preferences.execution to 'direct' and declare preferences.output_contract to get validated JSON records from a single call, routed to the cheapest model that holds the schema.
Schema valid · SHA-256 3f3c421778a3…
View input schema
{ "type": "object", "properties": { "workflow_id": { "type": "string", "description": "The workflow_id (wr_…) a plan call returned. Runs that stored plan without re-planning. Required when task_description is absent." }, "task_description": { "type": "string", "description": "Natural language description of the task, at least 2 words and at most 4,000 characters (200,000 across it and context.additional_context when task_type is compose). Be specific: include what you want produced, constraints, and context. Example: 'Design a caching layer for our API gateway with Redis integration.' Required when workflow_id is absent." }, "task_type": { "type": "string", "description": "Optional hint to bypass automatic classification. Passing it also removes the slowest classification tiers from the critical path, so send it whenever you know the shape. Multi-step deliverable types: solution_design, requirements_analysis, code_implementation, code_review, bug_fix, root_cause_analysis, incident_response, deployment_execution, competitive_scan, financial_analysis, research_task, documentation, governance_change, compliance_audit, data_security_assessment, performance_optimization, user_story_definition, implementation_prompt_generation. Atomic single-call types, which auto-select direct execution: tag, score, rerank, compose, extract_entities, parse_query, quick_research, quick_classification, quick_extraction, quick_scoring. GET /api/rigor/task-types lists the full vocabulary with each type's shape." }, "preferences": { "type": "object", "description": "Optional workflow preferences. The same shape on plan and execute; a plan stores them, and an execute by workflow_id inherits them.", "properties": { "rigor_level": { "type": "string", "enum": [ "quick", "standard", "thorough" ], "description": "Review depth. Default auto-detected from the task type. Controls analysis depth and cost." }, "tool_scope": { "type": "string", "enum": [ "mandatory", "recommended", "all" ], "description": "Which framework tiers to include." }, "model_preference": { "type": "string", "enum": [ "quality_first", "cost_optimized", "balanced" ], "description": "LLM model selection bias." }, "max_budget_usd": { "type": "number", "exclusiveMinimum": 0, "description": "Spending ceiling in USD. If the plan's quote is still above it after Rigor fits the plan (cheaper models, optional steps dropped), the plan returns a max_budget_exceeded warning and execute refuses with 402 max_budget_exceeded before anything runs or is charged." }, "require_approval": { "type": "boolean", "description": "Pause at pending_approval before the final step." }, "approval_before_step": { "type": "array", "items": { "type": "integer", "minimum": 0 }, "description": "Zero-based step indices where approval gates are inserted." }, "skip_frameworks": { "type": "array", "items": { "type": "string" }, "description": "Frameworks to exclude, by the display name the plan shows or by ID, any case. Unknown names are reported in warnings." }, "only_frameworks": { "type": "array", "items": { "type": "string" }, "description": "Restrict the plan to these frameworks (display name or ID). Applied before skip_frameworks; unknown names are reported in warnings, and if none resolves the request is refused with 400 UNKNOWN_FRAMEWORK." }, "add_frameworks": { "type": "array", "description": "Frameworks to inject into the sequence: a name, or { name, before_step } to choose the position.", "items": { "anyOf": [ { "type": "string" }, { "type": "object", "required": [ "name", "before_step" ], "properties": { "name": { "type": "string", "description": "Framework display name or ID." }, "before_step": { "type": "integer", "minimum": 0, "description": "Zero-based index into the filtered sequence before which to insert it." } } } ] } }, "use_connections": { "type": "boolean", "description": "Whether the plan may read your connected sources (a GitHub repository, an MCP server). false attaches none; true attaches all. Omitted, a source is attached only when the task is work on code or names a file, your own code, or the source itself." }, "execution": { "type": "string", "enum": [ "direct" ], "description": "Set to 'direct' to compose the plan's content frameworks into a single LLM call and route cost-first, using a per-task-type model floor that moves when a cheaper model earns the work. Research steps, process steps (classification-verify, review protocol, synthesis) and the quality review each remain separate calls, so this is not a one-call-per-workflow guarantee: for atomic task types, which have a single content framework, the call count matches standard execution and the saving is the model. Supplying output_contract replaces the quality-review call with deterministic validation, which is one fewer call. No intermediate outputs. Available at every tier. Auto-selected for atomic task types when no execution preference is given. Some atomic task types (for example cluster) run in direct mode only and are refused with 422 direct_mode_unavailable rather than falling back. Others fall back to standard execution when combined with require_approval or interactive mode, or when the plan exceeds the composition size limit. With attachments or prior_workflow_id, the workflow runs as standard execution instead, so the attachments and the earlier results are applied." }, "output_contract": { "type": "object", "description": "Only read when execution is \"direct\". Declares the JSON shape you want back, so the answer is generated against your schema (Anthropic forced tool-use, or an OpenAI-compatible json_schema response format) and validated against it before return, instead of returned as prose you have to parse. A conforming run also skips the quality-review call, costing 1 LLM call rather than 2. The schema is closed: a record carrying an undeclared key is rejected exactly like one missing a required key. Persists with the plan, so a later execute against the stored workflow_id inherits it. Not the same shape as the inference gateway's output_contract, which carries the items an answer must account for rather than a record schema.", "required": [ "task_type", "shape", "fields" ], "properties": { "task_type": { "type": "string", "description": "Your own label for the work. Echoed into telemetry. Not read as a framework name and does not change routing." }, "shape": { "type": "string", "enum": [ "object", "array" ], "description": "\"object\" for 1 record, \"array\" for 1 entry per input item." }, "fields": { "type": "array", "description": "The schema for 1 record.", "items": { "type": "object", "required": [ "name", "type" ], "properties": { "name": { "type": "string", "description": "The JSON key." }, "type": { "type": "string", "enum": [ "string", "number", "integer", "boolean", "array", "object" ], "description": "The value's type." }, "required": { "type": "boolean", "description": "Defaults to true. Set false for a field that may be absent." }, "enum": { "type": "array", "items": { "type": "string" }, "description": "Restricts a string field to a fixed set of values." }, "minimum": { "type": "number", "description": "Lower bound for a numeric field." }, "maximum": { "type": "number", "description": "Upper bound for a numeric field." }, "items": { "type": "object", "description": "The element schema, a field entry of this same shape. Only read when type is \"array\"." }, "properties": { "type": "array", "items": { "type": "object" }, "description": "The nested field entries, each of this same shape. Only read when type is \"object\"." }, "description": { "type": "string", "description": "Passed to the model as the field's description." } } } }, "count": { "type": "object", "description": "Bounds on the number of entries. Only read when shape is \"array\".", "properties": { "min": { "type": "number" }, "max": { "type": "number" } } }, "selection": { "type": "object", "description": "Use when you want the model to over-generate candidates and Rigor to sort, threshold, and cap them before the count bounds are checked.", "properties": { "scoreField": { "type": "string" }, "minScore": { "type": "number" }, "minCount": { "type": "number" }, "maxCount": { "type": "number" } } } } }, "independent_of": { "type": "object", "description": "Model creators and model ids that must not receive this workflow's content (for example when it grades output one model family produced). Covers every model call Rigor makes for the workflow: planning (classification, intent extraction, title), every step, research agents, review and quality checks, retrieval over files attached to the workflow, and per-step quality grading after the workflow ends. Not covered: processing when a file is uploaded (text extraction, audio and video transcription, indexing) and POST /api/rigor/enhance. A hard constraint: an excluded model, or one whose creator cannot be confirmed, is skipped; a step with no model left fails the workflow with policy_rule no_independent_model and is refunded; an optional planning or grading call with no model left is skipped. Not relaxed by any workspace policy enforcement mode. Fixed when the workflow is planned: an execute call on a stored plan that sends a different value is refused with 400; entry order does not matter.", "additionalProperties": false, "minProperties": 1, "properties": { "families": { "type": "array", "maxItems": 20, "items": { "type": "string", "pattern": "^[A-Za-z0-9][A-Za-z0-9._-]*$", "maxLength": 64 }, "description": "Model creators: anthropic, deepseek, google, minimax, mistral, moonshot, openai, qwen, xai or zhipu. Case does not matter; any other name is refused with 400." }, "models": { "type": "array", "maxItems": 20, "items": { "type": "string", "minLength": 1, "maxLength": 200 }, "description": "Model ids. A dated version or -latest alias of a listed id is excluded too." } } }, "cache_prefix_chars": { "type": "integer", "minimum": 0, "description": "How many leading characters of task_description stay the same across your calls. Direct execution marks that prefix for the provider's prompt cache, so repeated calls read it from cache. Absent or 0 declares nothing." } } }, "context": { "type": "object", "description": "Additional context for the workflow. Persisted with a plan, so an execute by workflow_id inherits it unless it sends its own.", "properties": { "additional_context": { "type": "string", "description": "Free-form context the workflow steps can reference, at most 16,000 characters (more for compose, see task_description)." } } }, "delivery": { "type": "object", "description": "How you receive progress and the result. Over REST the default is sse; over MCP it is polling, because most MCP clients cannot read a stream.", "properties": { "method": { "type": "string", "enum": [ "sse", "polling", "webhook" ], "description": "sse streams events; polling returns at once with poll_url; webhook posts the events in the webhooks section to webhook_url. Interactive delivery, which available_modes also lists, is chosen with mode, not here." }, "webhook_url": { "type": "string", "format": "uri", "maxLength": 2048, "description": "Required when method is webhook. HTTPS, resolving only to public addresses." } } }, "mode": { "type": "string", "enum": [ "full", "interactive" ], "description": "\"interactive\" returns interactive_waiting and runs 1 step per POST /api/rigor/step; otherwise \"full\"." }, "language": { "type": "string", "description": "BCP 47 language code for the deliverable (for example \"fr\", \"ja\", \"pt-BR\"). Overrides the organization default." }, "agent_id": { "type": "string", "maxLength": 200, "description": "Your own label for the calling agent, at most 200 characters. Recorded on the workflow and its cost rows; never used for routing or authorization." }, "prior_workflow_id": { "type": "string", "description": "Optional: the workflow_id (wr_…) of a finished workflow in your workspace to build on. Its results are passed to this run as context. It is charged as a new run at the normal quote." }, "context_fields": { "type": "array", "items": { "type": "string" }, "description": "Optional, with prior_workflow_id: the fields to carry over. Pass [\"full_step_results\"] for every step output in full." }, "refresh_connections": { "type": "boolean", "description": "Refresh your MCP connection state before the run starts. Defaults to true; a failed refresh never blocks the run." }, "workflow_ids": { "type": "array", "items": { "type": "string" }, "description": "Several planned workflow ids (from a multi-intent plan) to run without re-planning." } }, "additionalProperties": false }
Connections
streamable-http
https://plith.ai/api/mcp