Packages

I/O adapters for agent-doc frontmatter policy

Filesystem helpers for agent-doc

Content hashing helpers for agent-doc

Pure IPC protocol classification policy for agent-doc

TTL freshness primitives for agent-doc leases and sidecars

Lease-related process and filesystem adapters for agent-doc

Log timestamp and low-level ops-log formatting helpers for agent-doc

Structured markdown block AST for agent-doc session documents (tree-sitter-md based)

Phase 0 lossless-tree adapter for agent-doc markdown session documents (#lzlosstree): parses a document into a lazily LosslessTreeCrdt whose leaves own every byte, so render(parse(doc)) == doc. Shadow projection only — not an authority path.

Pure semantic memory ranking and event-shaping helpers for agent-doc

Pure merge policy functions for agent-doc documents

Harness-agnostic model tier resolution for agent-doc

Orchestration layer for agent-doc (routing, git, sessions, IPC, supervision, sync) — sits between the CLI and agent-doc-core

File-backed project configuration IO for agent-doc

Pure prompt-cache boundary and effectiveness policy for agent-doc

Prompt-cache history file I/O for agent-doc

Content-only live response TOC parsing and filtering for agent-doc

Secret redaction hygiene layer for agent-doc state and stream output

Pure session accretion decisions for agent-doc

Pure session registry mutation policy for agent-doc

SQLite-backed persistence layer for agent-doc (archive index, controller state store) — isolates the bundled-SQLite C build from agent-doc-orchestration recompiles

Pure editor visual syntax tokenization for agent-doc documents

Pure tmux realtime model state for agent-doc

Pure tmux command builders and parsers for agent-doc

Effectful tmux command execution adapter for agent-doc

Pure exchange topic-section parsing for agent-doc documents

Shared turn executor model vocabulary for agent-doc

Write command request types for agent-doc

An embedded agent for IoT devices.

Read-only local package and environment inspector

Environment daemon and protocol provider for Agent Foundation

Message/coordination server for AI coding agents using sqlitegraph pub/sub

Named counters that reset on epoch boundaries for per-session rate tracking

Composable error recovery strategies for LLM agent tool calls

Transport-neutral contracts for Agent Infra evaluation APIs

Vendored Evaluate v2 transport contract for agent-infra-sdk

Tiny in-process pub/sub for agent loop events. Sync-only Rust mirror of the Python agent-event-bus library: register handlers by event name (or `*` for every event), emit, and remove via opaque Subscription handles. Subscriber panics are isolated so the bus keeps dispatching.

Structured event emitter for agent runs. Append-only, JSON-line-serializable events with monotonic ids, run id, and timestamps. Zero deps beyond serde_json.

Structured lifecycle event log for AI agent runs

Non-interactive agent job runner. Runs commands as background jobs and returns structured JSON on stdout.

Execution contracts for Agent Infra

Version-gated experiments for coding-agent runtimes.

Ahp public package coordinate scaffold

Feature flags and capability gates for AI agent tool access

agent-fermata

A sandboxed HTTP client for AI agents with SSRF protection, domain policies, rate limiting, and resource controls

Agent File Tools — tree-sitter powered code analysis for AI agents

A firewall for AI coding agents: secret exfiltration, prompt injection, hidden-unicode, dangerous tool calls and egress control, as a CLI and an embeddable SDK

DEPRECATED — agent-first-config has been merged into agent-first-data. Use the `agent-first-data` crate (agent_first_data::document) and the `afdata` CLI instead. No further releases will be made here.

A naming convention that lets AI agents understand your data without being told what it means, plus a CLI and library for reading Markdown structure and safely editing structured JSON, TOML, YAML, dotenv, and INI documents.