Modules
AI integration layer for the Jido ecosystem.
Shared helper functions for Jido.AI skill actions.
A Jido.Action for chat-style LLM interactions with optional system prompts.
A Jido.Action for simple text completion without system prompts.
A Jido.Action for generating text embeddings using LLM embedding models.
A Jido.Action for generating structured JSON objects using LLM with schema validation.
A Jido.Action for breaking down complex goals into hierarchical sub-goals.
A Jido.Action for generating structured plans from goals.
A Jido.Action for prioritizing tasks based on given criteria.
Returns current quota usage and budget status.
Resets quota counters for a scope.
A Jido.Action for performing deep analysis of text/data with structured output.
A Jido.Action for getting clear explanations of complex topics.
A Jido.Action for drawing logical inferences from given premises.
Executes a reasoning strategy in an isolated runner agent context.
Clears retrieval memories in a namespace from the in-process store.
Recalls top-k memory snippets from the in-process retrieval store.
Upserts a memory snippet into the in-process retrieval store.
A Jido.Action for LLM calls with tool/function calling support.
A Jido.Action for direct tool execution without LLM involvement.
A Jido.Action for listing all available tools with their schemas.
Base macro for Adaptive strategy-powered agents.
Base macro for Jido.AI agents with ReAct strategy implied.
Base macro for Algorithm-of-Thoughts-powered agents.
Behavior for CLI adapters that drive different agent types.
Base macro for Chain-of-Draft-powered agents.
Base macro for Chain-of-Thought-powered agents.
Conversation context that accumulates messages for LLM projection.
A single entry in a conversation thread.
Directive to immediately emit a request error signal.
Directive to immediately emit a tool error result signal.
Helper functions for DirectiveExec implementations.
Directive asking the runtime to generate embeddings via ReqLLM.
Directive asking the runtime to generate an LLM response (non-streaming).
Directive asking the runtime to stream an LLM response via ReqLLM.
Directive to execute a Jido.Action as a tool.
Facade for effect policy, filtering, and state/directive application.
Shared helpers to normalize, filter, and apply effectful tool results.
Effect policy definition and filtering for tool-emitted effects.
Splode-based error handling for Jido.AI.
API-level errors from LLM providers
Authentication/authorization error
Rate limit exceeded error
Transient request failure error.
Sanitization helpers for user-safe error messaging.
Fallback error for unknown error types
Input/output validation errors
Input validation error
Structured output validation error
Base macro for Graph-of-Thoughts-powered agents.
Shared model alias resolution for public AI facades and ReAct runtime config.
AI observability boundary for telemetry events and spans.
Structured final-output contracts for Jido.AI runtimes.
Per-run FIFO queue for ReAct steering input.
Centralized default plugin composition for Jido.AI agent macros.
Conversational capability plugin with built-in tool calling support.
Cross-cutting model routing plugin.
A Jido.Plugin providing AI-powered planning capabilities.
Cross-cutting policy enforcement plugin for inbound AI/runtime signals.
Cross-cutting quota and budget enforcement plugin.
Plugin capability for isolated Adaptive strategy runs.
Plugin capability for isolated Algorithm-of-Thoughts runs.
Plugin capability for isolated Chain-of-Draft runs.
Plugin capability for isolated Chain-of-Thought runs.
Plugin capability for isolated Graph-of-Thoughts runs.
Plugin capability for isolated TRM runs.
Plugin capability for isolated Tree-of-Thoughts runs.
Cross-cutting retrieval and memory enrichment plugin.
Plugin that creates and manages a per-instance Task.Supervisor for Jido.AI agents.
Builds enriched user prompts with XML-tagged context sections.
Final stable quality checkpoint helpers.
Shared schema and helpers for text or multimodal user queries.
In-process quota counters backed by ETS.
Canonical namespace entrypoint for adaptive reasoning selection.
CLI adapter for Adaptive strategy agents.
Adaptive execution strategy that automatically selects the best reasoning approach.
Canonical namespace entrypoint for Algorithm-of-Thoughts reasoning.
CLI adapter for Algorithm-of-Thoughts-style agents.
Pure state machine for Algorithm-of-Thoughts (AoT) reasoning.
Canonical structured result contract for Algorithm-of-Thoughts (AoT) runs.
Algorithm-of-Thoughts (AoT) execution strategy for Jido agents.
Canonical namespace entrypoint for Chain-of-Draft reasoning.
CLI adapter for Chain-of-Draft-style agents.
Chain-of-Draft strategy implemented as a thin wrapper over delegated CoT runtime.
Canonical namespace entrypoint for Chain-of-Thought reasoning.
CLI adapter for Chain-of-Thought-style agents.
Pure state machine for the Chain-of-Thought (CoT) reasoning pattern.
Chain-of-Thought strategy delegated to an internal per-parent worker agent.
Canonical namespace entrypoint for Graph-of-Thoughts reasoning.
CLI adapter for Graph-of-Thoughts-style agents.
Pure state machine for the Graph-of-Thoughts (GoT) reasoning pattern.
Graph-of-Thoughts (GoT) execution strategy for Jido agents.
Helper functions for creating StateOps in Jido.AI strategies.
Public API for Task-based ReAct runtime.
Cancel an inactive ReAct checkpoint by issuing a new cancelled token.
Collect a terminal result from ReAct events or a checkpoint token.
Continue a ReAct runtime execution from a signed checkpoint token.
Start a ReAct runtime execution and return an event stream.
CLI adapter for Jido.AI.Agent modules (ReAct strategy implied).
Canonical configuration for the Task-based ReAct runtime.
Compatibility wrapper around Jido.AI.Runtime.Event.
Pending-input helpers for delegated ReAct runs.
Tracks a tool call in the ReAct runtime.
Behavior for advanced per-turn ReAct request shaping.
Task-based ReAct runner.
Signal envelope used by strategies/adapters to consume ReAct runtime events.
Runtime state for a single ReAct run.
ReAct strategy delegated to an internal per-parent worker agent.
Signed checkpoint tokens for caller-owned ReAct continuation.
Helpers for resolving request-scoped tool registries in ReAct.
Shared request lifecycle helpers for non-delegated reasoning strategies.
Canonical namespace entrypoint for Tiny-Recursive-Model reasoning.
Adaptive Computational Time (ACT) module for TRM strategy.
CLI adapter for TRM (Tiny-Recursive-Model) agents.
Shared utility functions for TRM (Tiny-Recursive-Model) modules.
Pure state machine for the TRM (Tiny-Recursive-Model) reasoning pattern.
Recursive Reasoning Engine for TRM (Tiny-Recursive-Model) strategy.
TRM (Tiny-Recursive-Model) execution strategy for Jido agents.
Deep Supervision Module for TRM (Tiny-Recursive-Model) strategy.
Canonical namespace entrypoint for Tree-of-Thoughts reasoning.
CLI adapter for Tree-of-Thoughts-style agents.
Pure state machine for the Tree-of-Thoughts (ToT) reasoning pattern.
Canonical structured result contract for Tree-of-Thoughts executions.
Tree-of-Thoughts (ToT) execution strategy for Jido agents.
Request tracking for AI agents with per-request isolation and correlation.
Represents a tracked request handle with correlation ID.
Request-scoped runtime event streaming helpers.
In-process retrieval memory store backed by ETS.
Canonical runtime event envelope shared across AI reasoning runtimes.
Signal for embedding generation completion.
Shared helpers for signal correlation and signal-safe payload shaping.
Signal for streaming LLM token chunks.
Signal for LLM streaming/call completion.
Signal for request lifecycle completion.
Signal for request rejection.
Signal for request lifecycle failure.
Signal for request lifecycle start.
Signal for tool execution completion.
Signal emitted when a tool execution starts.
Signal for token usage and cost tracking.
Unified skill abstraction for Jido agents.
First-class activation API for skills.
Tracks parsing and loading warnings for skill diagnostics.
Represents a single diagnostic warning.
Discovers skills from project-level and user-level skill directories.
Splode-based error handling for skill operations.
Skill not found in registry
Parse-level errors for SKILL.md files
Invalid YAML in frontmatter
No YAML frontmatter found in SKILL.md
Fallback error for unknown error types
Validation errors for skill specs
Invalid skill name format
Required field missing
Parses SKILL.md files into Jido.AI.Skill.Spec structs.
Renders skills into prompt text for agent system prompts.
ETS-backed registry for runtime-loaded skills.
Progressive disclosure of skill resources without eager loading.
Specification struct for skills, supporting both compile-time modules and runtime-loaded SKILL.md files.
Base macro for TRM (Tiny-Recursive-Model) agents.
Base macro for Tree-of-Thoughts-powered agents.
Adapts Jido Actions into ReqLLM.Tool structs for LLM consumption.
Canonical representation of a single LLM turn.
Helpers for merging provider usage metadata.
Validation utilities for Jido.AI runtime inputs.
Mix Tasks
Unified Jido AI CLI task.
Install and configure Jido AI for use in an application.
Maintainers only.
Manage and inspect skills from the command line.