Docs.rs

Operate Elastic Security as code with a safety-first CLI for security engineers.

Typed detection-rule model and endpoint wrappers for Elastic Security.

Core types, configuration, and transport for Elastic Security rule operations.

Every memory and recall backend for harness-rs behind one door: open_memory(url) / open_recall(url) pick the store from the URL. JSONL files with no features; SQLite FTS5, SurrealDB (embedded or remote, BM25 + HNSW hybrid) and CortexDB each behind a feature, so you compile only the store you run.

Multi-session serving core for harness-rs agents: pluggable auth, session store, per-request identity, and wired audit trail. Optional axum HTTP layer.

Document-extraction tool (read_document) for harness-rs — parse PDF/Word/Excel/PowerPoint locally in pure Rust, and fall back to an LLM for anything local parsing can't handle.

ReAct agent loop for harness-rs with measured guards: stuck detection, acceptance checks, a tool-result ceiling that spills oversized payloads to disk instead of truncating, per-call tool deadlines, subagent isolation, and session record/replay (JSONL).

LLM adapters for harness-rs — three protocol families via ApiKind::build(base_url, model, key): OpenAI-compatible, Anthropic Messages, Gemini native (Google Search grounding on by default). Plus embeddings, image, speech adapters and MockModel.

CortexDB-backed Memory for harness-rs: implements the harness-core Memory trait over CortexDB's MCP server or its native gRPC protocol (memory_save / memory_search), giving any harness agent semantic recall + a shared knowledge brain. Optional.

MCP client for harness-rs: expose a remote MCP server's tools as harness tools

Decapod is a Rust-built governance runtime for AI agents: repo-native state, enforced workflow, proof gates, safe coordination.

Rust agent framework based on harness-engineering principles (Lopopolo, Böckeler). Facade re-exporting the public surface of the harness-rs-* crates.

The tools a harness-rs agent reaches for — files, shell (with background jobs), memory, skills, tasks, dates, browser, recall, sensors, and an MCP server — in one crate. Tools whose dependencies are heavy live in their own crates instead.

Default World runtime (Tokio process runner, system clock, in-memory KV) and Context helpers for harness-rs.

Core traits and types for the harness-rs agent framework: Model, Tool, Guide, Sensor, Hook, Compactor, Skill, plus Context, World, Block, Event, FixPatch and the lifecycle events.