agentdir
01Cursor rule Cursor
Use AgentDir as invisible evidence capture for agentic engineering.
Cursor rule Cursor
Use AgentDir as invisible evidence capture for agentic engineering.
Instructions file GitHub Copilot
Instructions for jstxn/agentdir: AgentDir is agent-owned background instrumentation for this repository. The engineer installs or adopts it once; agents operate it during normal work.
Instructions file CodexOpenCode
Instructions for jstxn/agentdir: Use AgentDir as the local flight recorder for coding-agent work in this repository.
Instructions file
Instructions for jstxn/agentdir: AgentDir is agent-owned background instrumentation for this repository. The engineer installs or adopts it once; agents operate it during normal work.
Skill Claude CodeCodex
Use AgentDir as a local-first flight recorder for non-trivial coding-agent work in software repositories. Load at the start of coding tasks, before tests/lint/build/typecheck, and when asked about AgentDir evidence, audits, session replay, memory, context packs, adoption, or handoffs.
Command Claude Code
Run a regression-tests-first cleanup/refactor workflow to reduce slop.
Command Claude Code
Run one task through the full REIN flow from clarification through implementation, cleanup, review, and verification.
Command Claude Code
Understand a codebase and maintain .rein/codebase/ as a navigable map with evidence-backed findings.
Command Claude Code
Runtime-backed Socratic interview with durable state, clarity scoring, and spec bundle output.
Command Claude Code
Use after triage, before implementation. Breaks complex work into sequenced steps with risks, checkpoints, rollback boundaries, and interview-result handoff support.
Command Claude Code
Use after a miss, regression, or suspicious shortcut. Captures what failed and what rule should change.
Command Claude Code
Use before committing. Self-reviews the current diff for unrelated changes, debug leftovers, test weakening, and scope drift.
Command Claude Code
Use when requirements are too large, contradictory, or need negotiation. Produces a scope contract with what will and will not be done.
Command Claude Code
Use before ambiguous or multi-file work. Surfaces relevant files, assumptions, success criteria, and cheating boundaries.
Command Claude Code
Use before declaring code work complete. Runs verification, reviews diff hygiene, and emits an evidence report.
Skill Claude CodeCodex
Run a regression-tests-first cleanup/refactor workflow to reduce slop.
Skill Claude CodeCodex
Run one task through the full REIN flow from clarification through implementation, cleanup, review, and verification.
Skill Claude CodeCodex
Understand a codebase and maintain .rein/codebase/ as a navigable map with evidence-backed findings.
Skill Claude CodeCodex
Runtime-backed Socratic interview with durable state, clarity scoring, and spec bundle output.
Skill Claude CodeCodex
Use after triage, before implementation. Breaks complex work into sequenced steps with risks, checkpoints, rollback boundaries, and interview-result handoff support.
Skill Claude CodeCodex
Use after a miss, regression, or suspicious shortcut. Captures what failed and what rule should change.
Skill Claude CodeCodex
Use before committing. Self-reviews the current diff for unrelated changes, debug leftovers, test weakening, and scope drift.
Skill Claude CodeCodex
Use when requirements are too large, contradictory, or need negotiation. Produces a scope contract with what will and will not be done.
Skill Claude CodeCodex
Use before ambiguous or multi-file work. Surfaces relevant files, assumptions, success criteria, and cheating boundaries.