Prompt
Native Agent guidance.
Explore all 84 held-in training tasks used across the six formal domains. The matching 84 test tasks are reserved for final A0/A_last reporting and their trajectories are never exposed during evolution.
Tasks by domain
Compare the public training distribution by domain and effort tier. Hover or focus an axis to inspect its mix; click a label to filter the task registry below.
Held-in tasks
Loading the public task registry…
Test trajectories are never returned to this page.No public training tasks match these filters.
Editable capabilities
Prompt, Rules, Skills, Hooks, MCP, and Workflow are comparable across Pi, Claude Code, and Codex. Pi additionally exposes twelve physical source-code owners with clean rebuild and runtime-load verification.
Native Agent guidance.
Scoped behavioral constraints.
Reusable procedures and references.
Lifecycle-triggered scripts.
Typed external tool adapters.
Commands and subagent profiles.
Turns, tools, retry, cancel, and stop.
Runtime assembly and lifecycle.
Session orchestration and resume.
Tool results, errors, and truncation.
Provider-call context construction.
Summaries and overflow recovery.
Persistence and working-directory state.
Prompt discovery and assembly.
Skill discovery and precedence.
Read, edit, write, search, and shell.
Extensions, events, and dispatch.
Requests, streams, tools, usage, and errors.