Use this skill whenever the user mentions namba, namba help, namba project, namba regen, namba update, namba codex access, namba plan, namba harness, namba fix, namba run, namba queue, namba sync, namba pr, namba land, namba release, $namba-help, $namba-coach, $namba-create, $namba-queue, $namba-plan-review, $namba-review-resolve, $namba-release, or asks to use the Namba workflow.

Generated instruction contract for this repo skill:

Purpose: keep the role or command scope explicit, bounded, and testable.
Boundary: honor read-only versus mutating state effects, configured sandbox mode, and assigned file or workflow ownership.
Required output: report concrete actions, changed paths or artifacts, validation evidence, and pass/fail status or blockers.
Pass/fail criteria: claim success only when acceptance criteria and configured validation are satisfied; otherwise name the exact blocker and impact.
Evidence expectations: cite source artifacts such as SPEC files, .namba/ configs, diffs, test output, PR/check links, or generated manifests instead of relying on unsupported assertions.
Security responsibilities: never expose or commit secrets; treat auth, privacy, destructive commands, permission changes, and external network or credential use as security-sensitive.
Destructive command and escalation policy: do not run destructive commands unless explicitly requested; request approval for privileged, networked, or sandbox-blocked actions only when the active approval mode allows it, and otherwise report the blocker or use a safe non-escalating path.
Fallback implementer boundary: if a specialist path is unavailable and the main/default implementer takes over, stay within the assigned scope and preserve the same evidence and validation duties.
Portability: keep durable guidance non-project-specific unless the current repository config or SPEC explicitly provides the project detail.

Command mapping:

$namba-help: explain how to use NambaAI in this repository, which command or skill to use next, and where the authoritative docs live, without mutating repository state.
$namba-coach: restate the user's current goal, ask only essential routing questions when needed, correct clearly wrong command choices, and hand off to exactly one primary Namba workflow invocation without mutating repository state.
$namba-create: run the skill-first creation workflow for repo-local skills, project-scoped custom agents, or both. Keep the user-facing surface inside Codex and do not add a public namba create CLI command in this slice.
namba project: refresh repository docs and codemaps.
namba codex access: inspect or update Namba runner codex exec access defaults with explicit approval_policy / sandbox_mode flags; repo .codex/config.toml does not persist interactive approval or sandbox choices.
namba regen: regenerate AGENTS, repo-local skills, command-entry skills, Codex custom agents, readable role cards, and repo-local Codex config from .namba/config/sections/*.yaml.
namba update [--version vX.Y.Z]: self-update the installed namba binary from GitHub Release assets only after explicit user approval.
namba plan "<description>": create the next feature SPEC package under .namba/specs/, and when branch-per-work is enabled create or switch to the dedicated spec/... branch in the current workspace; unless --no-review is present, continue immediately with $namba-plan-review SPEC-XXX after creation.
$namba-plan-review: create or resolve a SPEC, run the three plan-review tracks in parallel when possible, and drive the advisory readiness loop before implementation starts.
namba harness "<description>": create the next harness-oriented SPEC package under .namba/specs/ through the same dedicated-branch planning contract for reusable agent, skill, workflow, or orchestration work.
$namba-plan-pm-review / $namba-plan-eng-review / $namba-plan-design-review: update product, engineering, or design review artifacts under .namba/specs/<SPEC>/reviews/ and refresh advisory readiness.
namba fix --command plan "<issue description>": create the next bugfix SPEC package under .namba/specs/ through the same dedicated-branch planning contract.
namba fix "<issue description>" or namba fix --command run "<issue description>": perform direct repair in the current workspace without creating a SPEC package.
$namba-review-resolve: resolve the target PR from the current branch when possible, inspect unresolved review threads with thread-aware GitHub state, record thread identity and outcome, validate before replying or resolving, include CI/check evidence when relevant, and avoid duplicating the configured review marker.
$namba-release: draft release notes from commits since the previous semver tag, write the notes to a durable per-version artifact, then hand off to the guarded namba release --version <version> --push path with a GitHub Release body that uses the generated notes.
namba run SPEC-XXX: execute the SPEC in the current Codex session. Read spec.md, plan.md, and acceptance.md, implement directly, validate, and sync artifacts.
namba run SPEC-XXX --solo|--team|--parallel: use the standalone CLI runner when you need explicit single-subagent, multi-subagent, or worktree-parallel execution semantics.
namba queue start <SPEC-RANGE|SPEC-LIST>: process already-existing SPEC packages one at a time through review, run, PR, checks, optional land, and local main refresh. Use status, resume, pause, and stop to operate the durable queue.
namba sync: refresh change summary, PR checklist, codemaps, advisory review readiness, and PR-ready docs after implementation.
namba pr "<title>": run sync plus validation by default, inspect PR checks, summarize bounded GitHub Actions failure snippets when checks fail, commit and push the current branch, and create or reuse a PR without requesting Codex review.
namba pr --review "<title>": perform the same PR handoff and explicitly request Codex review with the configured marker exactly once.
namba land: resolve the current branch PR, optionally wait for checks, merge when the PR is clean, and update local main safely.
namba doctor: verify that AGENTS, repo skills, .namba config, Codex CLI, and the global namba command are available.

Execution rules:

Treat .namba/ as the source of truth.
Prefer repo-local skills in .agents/skills/.
Prefer $namba-coach for read-only current-goal command coaching, prefer $namba-help for read-only usage guidance, and prefer command-entry skills such as $namba-create, $namba-run, $namba-queue, $namba-pr, $namba-land, $namba-release, $namba-plan, $namba-plan-review, $namba-harness, $namba-fix, $namba-review-resolve, $namba-plan-pm-review, $namba-plan-eng-review, $namba-plan-design-review, $namba-project, and $namba-sync when the user is invoking one Namba command directly.
Use the installed namba CLI for project, regen, update, codex access, plan, harness, fix, queue, pr, land, release, and sync when it will update repo state more reliably or self-update the installed CLI directly.
Keep .namba/specs/<SPEC>/reviews/*.md and readiness.md current when you use the plan-review workflow; the readiness summary is advisory unless the user explicitly asks for a gate.
For namba run in an interactive Codex session, prefer Codex-native in-session execution over recursively calling namba run, unless the user explicitly asks for standalone --solo, --team, --parallel, or --dry-run behavior.
Run validation commands from .namba/config/sections/quality.yaml before finishing.
Start each new SPEC or task on a dedicated work branch when .namba/config/sections/git-strategy.yaml enables branch-per-work collaboration.
Prepare PRs against main, write the title/body in Korean, and request Codex review only when namba pr --review or queue --review is explicit; use @codex review as the request command.