orient

Installs predicate enforcement into a project and authors its AGENTS.md goal-hierarchy.

10|2|Updated Jan 24, 2026
One-click install
npx skills add https://github.com/nrdxp/predicate --skill orient-nrdxp
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: orient
Source: https://github.com/nrdxp/predicate/tree/main/skills/orient
Command: npx skills add https://github.com/nrdxp/predicate --skill orient-nrdxp

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI coding agents drift from the true project goal over long sessions because they lack a persistent, structured anchor describing the project's goal, requirements, invariants, and unknowns. This Skill onboards a project into the predicate framework by installing enforcement hooks and authoring an AGENTS.md goal-hierarchy that every subsequent agent walk hydrates from. ## Core Features & Use Cases - Eight-phase orientation workflow: BOUND, ESTABLISH, MAP, SURVEY, ELICIT, NEST, AUTHOR, and RECONCILE phases map the repository, survey available tools, and draft contract-conformant AGENTS.md files, with two human seams for goal confirmation and reconciliation. - R/I/U hierarchy authoring: Generates Requirements, Invariants, and Unknowns sections with grounding and signposts, making AGENTS.md the persistent anchor that live context-map trackers hydrate from via Nickel contract validation. - Refresh and reorientation: Re-running performs a diff-scoped sync pass, and staleness triggers from the tracker_fresh.sh gate drive tactical re-validation or strategic escalation. - Use Case: When adopting predicate in a new repository, run the orientation workflow to install git hooks, initialize the .ledger sub-repository, map the codebase, and produce a root AGENTS.md that anchors all future agent walks. ## Quick Start Ask the agent to run the /orient workflow on your project to install predicate enforcement and author the AGENTS.md goal-hierarchy.

Frequently Asked Questions about orient

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I onboard a new project into predicate?

Run the /orient workflow, which executes eight phases: it composes a boundary IBC, installs git hooks via bootstrap/install.sh, maps the repository structure, surveys available skills and tools, then elicits the goal from a human before authoring AGENTS.md files. The workflow halts at two human seams for goal confirmation and reconciliation.

What is the AGENTS.md goal-hierarchy in predicate?

The AGENTS.md hierarchy is a persistent conditioning layer containing Requirements, Invariants, and Unknowns sections for the project and any components with distinct sub-goals. Every subsequent agent walk hydrates its live context-map tracker from these sections rather than reconstructing context from scratch.

Does re-running /orient overwrite my existing AGENTS.md?

Re-running performs a refresh pass where MAP and SURVEY are diff-scoped to changes since the last orientation. The RECONCILE phase presents only changed sections to the human, and the persistent layer mutates only through that human-approved reconciliation, never silently.

When should a component get its own AGENTS.md file?

Add a component AGENTS.md only when a directory has a goal genuinely distinct from and in service of the root goal. The default is root-only, since every additional file is a staleness surface that must be kept true and synced at each refresh.

What happens when the tracker_fresh.sh gate reports stale items?

A stale report triggers reorientation: tactical drift only updates the last_validated field in the context-map, while strategic drift (a fired signpost contradicting a requirement) escalates to a human and re-runs at minimum the MAP and ELICIT phases before writing back to AGENTS.md.