Skip to content

README.md

Latest commit

 

History

History
46 lines (37 loc) · 2.55 KB

README.md

File metadata and controls

46 lines (37 loc) · 2.55 KB
Raw

Agent

The agent runs one AI pipeline against a project. It takes resolved data in, reports through progress events, asks through an answerer you pass, and returns a result. It never reads a session, a store or a UI.

To call it from code, use runAgent. The developer interfaces cover it.

What goes in and out

  • In. A RunConfig with the caller's routing, a RunInput, and callbacks for progress and questions.
  • Out. One RunResult at the end.
  • Never in. A session, a store or the UI.

Where things live

What Where
The entry points index.ts and types.ts
The run types: config, input and result shared/types.ts
Progress events and the answerer progress.ts
Sequences, harnesses and route resolution runner
The wizard tools both harnesses share tools
The benchmark pipeline middleware
YARA scans of tool calls (Pre/PostToolUse) and of installed skills yara-hooks.ts

Import runtime values from @agent and types from @agent/types. @agent exports runAgent, RunOutcome (from @shared/run-state, so the hosts read it without the agent), AgentSignals, WIZARD_TOOL_NAMES, DEFAULT_BINDING, scanVerdict, scanInstalledSkill and streamMcpPrompt, the MCP tutorial's prompt stream. pnpm lint rejects a deeper import from any other layer, through ESLint no-restricted-syntax. Files inside src/agent import each other by relative path. Nothing enforces this: the agent's project maps @agent, @agent/types and @agent/* through tsconfig.no-tui.json, and its tests use them. The agent itself may import @env, @shared/* and @utils/*, and nothing else; see layer boundaries.

Tools

tools/mcp.ts lists the wizard tools, and tools/tools.ts implements them. Pi registers the same tools in harness/pi/tools.ts, and reaches the PostHog MCP server through harness/pi/mcp.ts.