recall-scoutlisted

<!-- Generated by scripts/build-codex-plugin.ts; do not edit. --> ## Codex routing Use the plugin MCP tools `memory_search`, `memory_hybrid_search`, and `context_for_agent` for the memory-first steps; keep the workflow sensitive-data boundary intact. The canonical workflow below remains authoritative, but equivalent behavior is not assumed across hosts. Produce a **scout report** for the current repository: a fast, structured orientation for whoever is about to work in this codebase. An optional focus argument narrows the scout to a subsystem, path, or question (e.g. `auth`, `the extraction pipeline`, `src/db`). With no focus, scout the repo as a whole. This is the canonical scouting workflow; platform guides reference it and only remap tool names. Do not hand-copy these steps into `FOR_CLAUDE.md`, `FOR_PI.md`, `FOR_OPENCODE.md`, or `opencode/recall-memory.md` — those files point here and supply only their own tool-name mapping. ## Live context Three bounded orienters, each well under 1 KB of output. On Claude Code the `!` prefix inlines their output here at prompt load; on hosts without the `!` mechanic, run the same three commands manually as your first step. - Recent commits: !`git log -10 --oneline` - Working tree: !`git status -s | head` - CodeGraph probe: !`codegraph status --json` ## Workflow 1. **Memory first.** Before reading git history or source files, search Recall for what is already known about this repo and the focus. Use keyword search for names and