Provides the foundational value objects and adapter contracts for tool-call execution in ask-rb: ToolCall, ToolResult, ExecutionContext, and the ToolExecutor adapter interface.

Isolated code execution via Local (subprocess + rlimits), Docker (containers), Daytona (remote sandboxes), and Cloudflare (Workers sandbox). Zero external runtime dependencies.

A compact Ruby DSL for building standards-oriented JSON Schema documents. Zero dependencies.

Error context for agents via Sentry for the ask-rb ecosystem.

Provides immutable session records, event envelopes, an in-memory store with concurrency guardrails, a durable provider-backed store for restart-safe sessions, a replayable host with subscriptions, a runtime tool-lifecycle sink, and a state reducer that reconstructs session snapshots from event history.

The contract between the ask session host and every client (terminal TUI, web console, bots, IDE) — plus the shared Client and Host. Defines the canonical session event vocabulary, the resolvable interaction types (approvals, plan proposals, user input), the client-to-host method surface, protocol versioning, and generated JSON Schema artifacts. No runtime, no transports — just the contract.

Discovers, validates, and formats agent skills from project directories, user config, and installed gems. Ships built-in skills (codebase exploration, debugging methodology). Each skill is a markdown file with step-by-step instructions that the agent loads on demand.

Provides authenticated Slack client helper, context metadata, and error guide for AI agents.

Error context for agents via SolidErrors for the ask-rb ecosystem.

Provides Ask::State adapter implementations for SQLite, Redis, PostgreSQL, and MySQL. Each backend implements the Ask::State::Adapter contract (get/set/delete, distributed locking, message queues, and ordered lists) so agents, sessions, and workflows pick the right persistence layer with one line of code.

The first thin client of the ask ecosystem: an interactive terminal coding agent that speaks the canonical ask-session-protocol and nothing else. Spawns the host (ask-app-server) over stdio or attaches to a running host over its unix socket — the same protocol any other client (web console, bots) speaks, so multiple clients can share the same live sessions.

A Rails-free foundation for token-based usage tracking. Count real tokens with tiktoken, price them at any rate, declare activity costs, and run a wallet/ledger engine against a pluggable store. Pair with ask-tokens-rails for ActiveRecord persistence.

Adds an ActiveRecord-backed wallet store, has_token_wallet concern, token wallet/transaction models, an install generator with migrations, and an expiry sweep job to any Rails app using ask-tokens.

A Rails-free foundation for token-based usage tracking. Count real tokens with tiktoken, price them per million, declare activity costs, and run a wallet/ledger engine against a pluggable store. Pair with ask-token-usage-rails for ActiveRecord persistence.

Adds an ActiveRecord-backed wallet store, has_token_wallet concern, token wallet/transaction models, an install generator with migrations, and an expiry sweep job to any Rails app using ask-token-usage.

Defines Ask::Tool (base class), tool discovery, and the Ask::Result returned by tools.

Bash, Read, Write, Edit, Glob, Grep, and Code tools for the ask-rb ecosystem.

Fetches a URL and converts its content to clean markdown for LLM consumption. A pluggable backend chain: pure Ruby httpx + Nokogiri + reverse_markdown by default, a Jina Reader fallback for JS-rendered or blocked pages, and a real-Chrome fallback (Ferrum) that renders JavaScript and lets auto-solving Cloudflare challenges complete. The capability layer only — tool framing (MCP servers, agent tools) is provided by the consumers.

A minimal MCP (Model Context Protocol) server that exposes ask_web_fetch as a callable tool over stdio. Designed for use with clients that support MCP (ZCode, Claude Code, etc.), it fetches a URL and returns clean markdown through the ask-web-fetch backend chain — fast pure-Ruby httpx fetch first, a real Chrome (launched or CDP-attached) for JS-rendered and challenge-gated pages, with Jina Reader and self-hosted Crawl4AI in between. Terminal failures (parked domains, empty pages, dead 4xx) surface as their deterministic error class, so clients never retry the unretryable. The tool shell (name, schema, call) lives here, wrapping the Ask::WebFetch library.

Web search returning results as clean numbered markdown. Two selectable backends: a local SearXNG instance (SEARXNG_URL, the default path for everyone) and TinyFish's hosted API (opt-in: SEARCH_BACKEND=tinyfish plus a TINYFISH_API_KEY or ask-auth credentials key). The capability layer (WebSearch.search); the native Ask::Tools::WebSearch agent tool is an optional integration that registers when ask-tools is present.