Agent Context (wh prime)

wh prime outputs a complete CLI context dump optimized for AI agents. It’s designed to be the first command an agent runs at session start or after context compaction.

Usage

# Markdown output (human-readable, ~2,000 tokens)
wh prime

# Structured JSON output
wh prime --json

When to Use

• Session start — give the agent full CLI context before it begins work
• After compaction — restore context that was dropped during conversation compression
• New agent — bootstrap a fresh agent with everything it needs to operate

Markdown Output

The default output includes:

• Environment — default repo (from WARMHUB_REPO)
• Core concepts — repo, thing, assertion, shape, commit, wref definitions
• Wref quick reference — local/canonical forms, version modifiers, batch tokens
• Command reference — all domains and verbs with args and aliases
• Common workflows — explore, create, modify, query patterns
• Agent tips — --json, --live, prefix expansion, fuzzy matching

Example excerpt:

## Core Concepts
- **Repo**: org/repo namespace containing things, assertions, shapes, commits
- **Thing**: A named entity (e.g., `player`, `room`) versioned through commits
- **Assertion**: A claim about a thing with shape-validated data
- **Shape**: A schema defining the data structure for things/assertions
- **Commit**: An atomic batch of add/revise operations
- **wref**: A human-readable reference to a thing (its name)

JSON Output

The --json output includes:

{
  "version": "0.1.0",
  "defaultRepo": "myorg/myrepo",
  "defaultOrg": "myorg",
  "wrefSyntax": {
    "localExamples": ["GameState", "GameState@v3", "Game/player"],
    "canonicalFormat": "wh:org/repo/Shape/name",
    "versionModifiers": ["@HEAD", "@vN", "@ALL"],
    "batchTokens": { "allocate": "$N", "reference": "#N" }
  },
  "domains": [
    {
      "domain": "org",
      "summary": "Organization management",
      "verbs": [
        { "name": "create", "summary": "Create a new organization", "args": "<name>", "status": "live" }
      ]
    }
  ]
}

The JSON format is structured for programmatic consumption — each domain lists its verbs with args, aliases, flags, and status.

Token Budget

Prime output is kept compact (~2,000 tokens) to leave room in the agent’s context window for reasoning and conversation.

MCP Equivalent

For MCP-connected agents, the equivalent bootstrapping tool is warmhub_repo_describe. See MCP Quickstart for usage and MCP Tools Reference for the full schema.

Best Practice

Agents should call wh prime (CLI) or warmhub_repo_describe (MCP) as their first action, then use targeted tools for specific operations. This avoids guessing at schema details and gives the agent accurate write contracts from the start.