@jstn-sdk/rcs 0.1.0

package/docs/release-notes-v0.1.0.md

@@ -0,0 +1,53 @@
+> Archive context: this canonical release note consolidates pre-v0.1.0 retained release-note collateral into one RCS-owned historical surface for the v0.1.0 ship line.
+
+# Roblox Creator Skills v0.1.0
+
+## Summary
+
+`v0.1.0` is the first strict RCS-owned ship candidate after the fork-to-RCS transition. This release consolidates the retained pre-v0.1.0 release-note train into one canonical archival note and defines the current shipped product surface as **Roblox Creator Skills for Codex**.
+
+## Canonical product surface
+
+- Package: `@jstn-sdk/rcs`
+- Owner: `JustineDevs`
+- Repository: `https://github.com/JustineDevs/roblox-ai-os`
+- Product: `Roblox Creator Skills`
+- Short identity: `RCS`
+
+## Shipped product direction
+
+- Creator-first workflow family:
+  - `$brief`
+  - `$blueprint`
+  - `$forge`
+  - `$crew`
+  - `$autoforge`
+- Psychology design layer for Roblox experience design:
+  - audience, motivation, loop, retention, social, progression, status, mastery, community, reward-loop, daily-loop, event-loop, and FOMO surfaces
+- Roblox pre-action gate:
+  - resource grounding
+  - official-term normalization
+  - architecture-first modular planning
+  - explicit `PRE_ACTION_COMPLETE` before implementation
+- First-party Codex plugin bundle:
+  - mirrored skills
+  - plugin metadata
+  - bundled psychology and Roblox planning assets
+
+## Consolidation note
+
+The old retained release-note files from the pre-v0.1.0 migration train were preserved only as transitional collateral. They are now consolidated into this single archival surface to avoid version drift, duplicated wording, and mixed ownership language across release/readiness docs.
+
+## Historical scope covered by the consolidation
+
+Per-version `docs/release-notes-0.*.md` files from earlier trains have been **removed from the tree**; their substance is represented here at a high level and remains recoverable from **git history** if a line-by-line comparison is needed.
+
+## Verification and release-collateral expectations
+
+- **`CHANGELOG.md`** is the canonical changelog surface (currently the consolidated **`[0.1.0]`** entry plus `[Unreleased]`).
+- **`RELEASE_BODY.md`** is the release-body / GitHub summary template aligned with that ship line.
+- **`docs/qa/release-readiness-*.md`** are retained readiness snapshots; update or archive them if they still point at removed per-version release-note paths.
+
+## Archival boundary
+
+This file is archival collateral, not a claim that every retained pre-v0.1.0 development snapshot is independently ship-ready. It exists to keep one coherent historical release-note surface while the canonical shipped identity remains RCS-owned and Roblox-focused.

package/docs/reports/macos-m1-high-cpu-usage-2026-04-16.md

@@ -0,0 +1,78 @@
+> Archive context: retained release and migration collateral from the pre-v0.1.0 fork-to-RCS transition. Links are normalized to the canonical JustineDevs repository, but historical chronology and contributor references may still describe pre-fork development.
+
+## macOS M1 High CPU Usage Investigation
+
+Date: 2026-04-16
+
+### Symptom
+
+Launching `rcs --madmax --xhigh` on an Apple Silicon Mac caused `syspolicyd`
+to spike, often alongside visible CPU churn in `sysmond`.
+
+### Root Cause
+
+The dominant trigger was not `syspolicyd` itself. RCS startup entered the
+fallback watcher path in a repository with stale runtime state, which caused
+`leader_nudge` polling to run roughly every 250 to 350 ms.
+
+That polling path called `isLeaderRuntimeStale()`, which read leader git
+activity by shelling out to multiple short-lived `git` processes:
+
+- `git rev-parse --git-dir`
+- `git symbolic-ref --quiet --short HEAD`
+- `git rev-parse --git-path logs/HEAD`
+- `git rev-parse --git-path logs/refs/heads/<branch>`
+- `git show -s --format=%ct HEAD`
+
+On macOS, each of those short-lived `git` execs triggered code-signing and
+policy checks in `syspolicyd`. The repeated exec bursts were sufficient to
+drive `syspolicyd` CPU well above normal idle levels during startup.
+
+### Patch
+
+File changed:
+
+- `src/team/leader-activity.ts`
+
+Implementation:
+
+- Added a process-local cache for leader git activity lookups.
+- Cache key is the repository root derived from `stateDir`.
+- Cache TTL is bounded between `1000 ms` and `5000 ms`.
+- TTL is computed as `thresholdMs / 4`, clamped to that range.
+- `readLeaderRuntimeSignalStatuses()` now uses the cached git activity lookup
+  instead of invoking the git-reading path on every poll cycle.
+
+### Why This Shape
+
+- The fix targets the verified hotspot directly.
+- It preserves stale-detection semantics.
+- It avoids broader changes to watcher cadence or team lifecycle behavior.
+- It keeps the diff small and reversible.
+
+### Validation
+
+Controlled reproduction before the patch:
+
+- `rcs --madmax --xhigh` in `GhostVMPriv`
+- `notify-fallback-watcher` active with `poll_ms = 250`
+- `leader_nudge` firing repeatedly
+- `syspolicyd` observed around `66% CPU`
+- `fs_usage -f exec` showed repeated `git` exec bursts
+
+Controlled reproduction after the patch:
+
+- Same repository and startup command
+- Same `leader_nudge` loop still active
+- 4-second post-start sampling window observed `0` `git` execs
+- `syspolicyd` dropped to low single-digit CPU usage
+
+### Remaining Risks
+
+- Stale `.rcs/state` can still keep `leader_nudge` polling active; this patch
+  removes the dominant git-exec hotspot but does not stop the poll loop.
+- Other synchronous git metadata reads still exist in separate paths such as
+  operational event enrichment and HUD rendering, but they were no longer the
+  dominant startup hotspot in the reproduced window.
+- Git activity freshness is now cached briefly, so stale detection can lag by
+  at most the configured cache TTL.

package/docs/reports/open-prs-dev-readiness-2026-04-09.md

@@ -0,0 +1,45 @@
+> Archive context: retained release and migration collateral from the pre-v0.1.0 fork-to-RCS transition. Links are normalized to the canonical JustineDevs repository, but historical chronology and contributor references may still describe pre-fork development.
+
+# Open PR readiness matrix for `dev` — 2026-04-09 UTC
+
+Current `dev` head reviewed: `8656d21149a0772369f697df002f7bf85002db8d`.
+
+## `dev` CI note
+
+Push run `24141814673` on `dev` finished red because long jobs were cancelled, not because the fast validation jobs failed.
+
+- Passed on `dev`: Rust Format, Rust Clippy, Lint, Typecheck (Node 20/22), Test (Node 22 / smoke), Coverage Report (Rust), Coverage Gate (Team Critical), Ralph Persistence Gate.
+- Cancelled on `dev`: `Test (Node 20 / full)`, `Coverage Report (TypeScript Full)`.
+- Failed gate: `CI Status`.
+- Workflow reason: `.github/workflows/ci.yml:260-289` requires every `needs.*.result` to equal `success`; cancelled/skipped results force the gate red.
+
+## Readiness matrix
+
+| PR | Summary | Current checks | Merge blockers | Recommendation |
+| --- | --- | --- | --- | --- |
+| #1403 | Shared-session shutdown hardening + regression coverage | Fast checks mostly green, but both Typecheck jobs failed in run `24173665271`; full Node 20 suite still running | **Hard blocker:** `npm run check:no-unused` fails locally on PR head `e802eff1` with `src/team/runtime.ts(20,3): error TS6133: 'isNativeWindows' is declared but its value is never read.` | **Do not merge yet.** Fix unused import, rerun PR CI, then reassess. |
+| #1402 | tmux HUD self-healing / prompt-submit reconcile | Lint, typecheck, smoke, rust coverage, team coverage gate passed; full Node 20 and TS full coverage still pending | Pending long jobs; touches `src/cli`, `src/hud`, and native-hook operational events, so I would not merge before `dev` CI is green | **Good candidate once long jobs finish green and `dev` is green.** Recommended ahead of docs-only work. |
+| #1400 | Unknown `$token` duplicate continuation + stale Ralph state fix | Lint, typecheck, smoke, rust coverage, team coverage gate passed; full Node 20 and TS full coverage still pending | Pending long jobs; overlaps `src/scripts/codex-native-hook.ts` and its tests with #1380 | **Merge after #1380 is in `dev`, then rebase/retest.** Otherwise likely conflict/re-review churn. |
+| #1382 | Stale team worktree cleanup at startup | PR is open, but its head commit `e5f9ffe7...` is already an ancestor of current `dev` head | No merge work left; CI Status failure is stale noise from cancelled long jobs on an already-landed change | **Close as stale/already merged.** No merge needed. |
+| #1380 | Stale deep-interview stop-hook gating fix | Fast checks passed; CI Status failed only because long jobs were cancelled in old run `24142611197` | Long jobs were cancelled; overlaps same native-hook files as #1400 | **Best first merge among the hook fixes once `dev` CI is green.** Land before #1400. |
+| #1357 | AGENTS/template token reduction | Fast checks passed; CI Status failed only because long jobs were cancelled in old run `24145469951` | Long jobs cancelled; PR body/checklist says only two files changed, but diff also changes `CLAUDE.md` | **Hold for manual review after runtime fixes.** Lower urgency than #1380/#1400/#1402. |
+
+## Recommended merge order
+
+1. Close **#1382** as already merged into `dev`.
+2. After `dev` CI is rerun/green, merge **#1380**.
+3. Rebase/retest and merge **#1400**.
+4. Merge **#1402** once its pending jobs finish green.
+5. Re-review **#1357** (body/diff mismatch, prompt-contract risk) before merging.
+6. Fix **#1403** unused import + rerun CI before considering merge.
+
+## Evidence collected
+
+- `gh run view 24141814673 --json ...` for current `dev` CI failure shape.
+- `gh pr checks 1402`, `1400`, `1382`, `1380`, `1357`, `1403` for current PR statuses.
+- `gh pr diff --name-only` for overlap checks.
+- `git rev-list --left-right --count dev...pr-<n>` and `git merge-base dev pr-1382` for ancestry / already-merged detection.
+- Local PR verification for #1403 in `/tmp/worker3-pr1403`:
+  - `npm ci`
+  - `npm run build`
+  - `npm run check:no-unused` → unused import failure in `src/team/runtime.ts`.

package/docs/runtime-model-metadata-cleanup.md

@@ -0,0 +1,37 @@
+# Runtime Model Metadata Cleanup
+
+## Why
+
+Benchmark runs showed spawned child agents being labeled with legacy model aliases such as `sonnet`, even when RCS was launched under Codex with GPT-5.x reasoning settings.
+
+The prompt/skill layer was already sanitized, but runtime metadata still used historical aliases in `src/agents/definitions.ts`. Those aliases were also used when generating native agent configs, which made legacy labels leak into spawned-agent UX and benchmark output.
+
+## What changed
+
+### 1. Agent runtime metadata
+- Replaced `model: haiku|sonnet|opus` with `reasoningEffort: low|medium|high` in `src/agents/definitions.ts`.
+- Kept posture and model-class concepts intact.
+
+### 2. Native agent config generation
+- Updated `src/agents/native-config.ts` to derive `model_reasoning_effort` directly from `agent.reasoningEffort`.
+- Removed the legacy alias translation table.
+
+### 3. Tests
+- Updated `src/agents/__tests__/definitions.test.ts` and `src/agents/__tests__/native-config.test.ts` to match the new runtime metadata shape.
+- Existing prompt/skill sanitization test remains as a guard against old aliases reappearing in prompt content.
+
+### 4. Supporting cleanup
+- Updated `src/verification/verifier.ts` comment wording from legacy alias names to low/medium/high reasoning wording.
+
+## Expected effect
+
+After rebuilding and rerunning `rcs setup --force`, spawned-agent metadata should stop surfacing `haiku` / `sonnet` / `opus` as if they were active runtime models for Codex-based RCS runs.
+
+## Re-test steps
+
+```bash
+npm run build   # TypeScript build
+node bin/rcs.js setup --scope project --force
+```
+
+Then rerun the fork benchmark and watch for any remaining `Model: sonnet` lines. If they still appear after this cleanup, the remaining source is likely outside prompt/skill/runtime metadata generation and should be investigated in Codex display integration or cached/generated config artifacts.

package/docs/shared/agent-tiers.md

@@ -0,0 +1,56 @@
+# Agent Tiers
+
+This file defines practical tier guidance for RCS agent routing.
+
+## Mental Model
+
+RCS now separates three concepts:
+
+- `role`: what the agent is responsible for (`executor`, `planner`, `architect`)
+- `tier`: how much reasoning/cost to spend (`LOW`, `STANDARD`, `THOROUGH`)
+- `posture`: how the role behaves (`frontier-orchestrator`, `deep-worker`, `fast-lane`)
+
+Use role to choose responsibility, tier to choose depth, and posture to choose operating style.
+
+## Tiers
+
+- `LOW`:
+  Fast lookups and narrow checks.
+  Use for simple exploration, style checks, and lightweight doc edits.
+  Typical roles: `explore`, `style-reviewer`, `writer`.
+
+- `STANDARD`:
+  Default tier for implementation, debugging, and normal verification.
+  Typical roles: `executor`, `debugger`, `test-engineer`, `quality-reviewer`.
+
+- `THOROUGH`:
+  Use for architectural, security-sensitive, or high-impact multi-file work.
+  Typical roles: `architect`, `critic`, `security-reviewer`, `executor`.
+  Note: `deep-executor` is deprecated; route implementation to `executor`.
+
+## Selection Rules
+
+1. Start at `STANDARD` for most code changes.
+2. Use `LOW` only when the task is bounded and non-invasive.
+3. Escalate to `THOROUGH` for:
+   - security/auth/trust-boundary changes
+   - architectural decisions with system-wide impact
+   - large refactors across many files
+4. For Ralph completion checks, use at least `STANDARD` architect verification.
+
+## Posture Guidance
+
+- `frontier-orchestrator`:
+  - Best for steerable frontier models and leader-style roles.
+  - Prioritizes intent classification, delegation, verification, and architectural judgment.
+  - Typical roles: `planner`, `analyst`, `architect`, `critic`, `code-reviewer`.
+
+- `deep-worker`:
+  - Best for implementation-heavy roles that should carry work to completion.
+  - Prioritizes direct execution, minimal diffs, and strict verification.
+  - Typical roles: `executor`, `debugger`, `test-engineer`, `build-fixer`.
+
+- `fast-lane`:
+  - Best for cheap/fast models used for triage, search, and narrow synthesis.
+  - Prioritizes quick routing, concise search, and escalation over deep autonomous work.
+  - Typical roles: `explore`, `writer`, and lightweight research/search specialists.

package/docs/skills.html

@@ -0,0 +1,132 @@
+<!doctype html>
+<html lang="en">
+  <head>
+    <meta charset="UTF-8" />
+    <meta name="viewport" content="width=device-width, initial-scale=1.0" />
+    <title>Skills - RCS Docs</title>
+    <link rel="stylesheet" href="./style.css" />
+  </head>
+  <body>
+    <header>
+      <nav class="container nav" aria-label="Primary">
+        <a class="brand" href="./index.html">RCS</a>
+        <a href="./index.html">Home</a>
+        <a href="./getting-started.html">Getting Started</a>
+        <a href="./agents.html">Agents</a>
+        <a class="active" href="./skills.html">Skills</a>
+        <a href="./integrations.html">Integrations</a>
+      </nav>
+    </header>
+
+    <main class="container">
+      <h1>Skills Reference</h1>
+      <p class="muted">Canonical creator workflow triggers, psychology-driven design skills, and installable support surfaces.</p>
+
+      <h2>Creator Workflow</h2>
+      <ol>
+        <li><code>$brief</code> - capture the request, goals, constraints, audience, genre, and outcome.</li>
+        <li><code>$blueprint</code> - generate structured design and planning artifacts.</li>
+        <li><code>$forge</code> - generate concrete creator-facing deliverables and system specs.</li>
+        <li><code>$crew</code> - orchestrate multi-role creator workflows.</li>
+        <li><code>$autoforge</code> - run an end-to-end creator workflow with minimal supervision.</li>
+      </ol>
+
+      <h2>Roblox Pre-Action Gate</h2>
+      <ul>
+        <li>Before Roblox code generation, RCS must gather references, synthesize understanding, standardize terms, and design the modular file tree first.</li>
+        <li><code>PRE_ACTION_COMPLETE</code> must remain false until the pre-action artifact is complete.</li>
+        <li>Use <code>docs/reference/roblox-pre-action-protocol.md</code> and <code>templates/roblox/pre-action-plan.md</code> as the required planning surfaces.</li>
+      </ul>
+
+      <h2>Psychology Design</h2>
+      <ul>
+        <li><code>$brief:audience</code> - identify target fantasy, habits, desire language, session length, and return motive.</li>
+        <li><code>$brief:motivation</code> - rank progression, status, FOMO, mastery, and community by genre fit and risk.</li>
+        <li><code>$blueprint:psych</code> - generate the player psychology blueprint from desire backward.</li>
+        <li><code>$blueprint:loop</code> - define trigger → action → reward → repeat across session, daily, weekly, and comeback loops.</li>
+        <li><code>$blueprint:retention</code> - define D1/D7/D30 assumptions, cadence, and comeback hooks.</li>
+        <li><code>$blueprint:social</code> - define how the experience becomes a social machine.</li>
+        <li><code>$forge:reward-loop</code> - generate trigger/action/reward/escalation/reset/fail-state specs.</li>
+        <li><code>$forge:daily-loop</code> - generate daily return motivators and bounded streak/event logic.</li>
+        <li><code>$forge:event-loop</code> - generate limited-time event structures with urgency and re-entry logic.</li>
+        <li><code>$forge:progression</code> - generate XP, level, multiplier, prestige, rebirth, and unlock ladders.</li>
+        <li><code>$forge:status</code> - generate prestige and flex systems with visible milestones.</li>
+        <li><code>$forge:fomo</code> - generate urgency mechanics with fairness notes and anti-frustration guardrails.</li>
+        <li><code>$forge:mastery</code> - generate learning, optimization, discovery, and skill-expression systems.</li>
+        <li><code>$forge:community</code> - generate systems that make playing with others sticky.</li>
+      </ul>
+
+      <h2>Canonical Framework</h2>
+      <ul>
+        <li>Design from player desire backward.</li>
+        <li>Use the five drivers: progression, status, FOMO, mastery, community.</li>
+        <li>Map each driver through mechanic patterns, UI feedback, economy impact, retention impact, monetization fit, risk notes, and Roblox-fit notes.</li>
+      </ul>
+
+      <h2>Compatibility Runtime Layers</h2>
+      <p class="muted">These are advanced lower-level execution layers. They remain available for compatibility, diagnostics, and operator control, but they are not the primary creator workflow.</p>
+      <ul>
+        <li><code>$team</code> - lower-level coordinated team runtime retained for compatibility and operational control.</li>
+        <li><code>$ralph</code> - lower-level persistence loop retained for compatibility and runtime control.</li>
+        <li><code>$autopilot</code> - lower-level autonomous execution runtime retained for compatibility.</li>
+        <li><code>$ultrawork</code> - lower-level max-parallel execution mode retained for compatibility.</li>
+        <li><code>$visual-verdict</code> - structured visual QA loop for screenshot/reference matching</li>
+        <li><code>$ecomode</code> - token-efficient model routing</li>
+        <li><code>$ultraqa</code> - test/verify/fix loop</li>
+        <li><code>$swarm</code> - team compatibility facade</li>
+      </ul>
+
+      <h2>Planning & Clarification</h2>
+      <ul>
+        <li><code>$brief</code> - canonical briefing surface</li>
+        <li><code>$blueprint</code> - canonical planning surface</li>
+        <li><code>$plan</code> - supporting strategic planning when you are not using the canonical creator path</li>
+        <li><code>$deep-interview</code> and <code>$ralplan</code> remain available as compatibility/internal planning layers.</li>
+      </ul>
+
+      <h2>Agent Shortcuts</h2>
+      <ul>
+        <li><code>$analyze</code></li>
+        <li><code>$deepsearch</code></li>
+        <li><code>$tdd</code></li>
+        <li><code>$build-fix</code></li>
+        <li><code>$code-review</code></li>
+        <li><code>$security-review</code></li>
+        <li><code>$visual-ralph</code> (advanced compatibility visual runtime)</li>
+        <li><code>$frontend-ui-ux</code></li>
+        <li><code>$git-master</code></li>
+        <li><code>$review</code></li>
+      </ul>
+
+      <h2>Utilities</h2>
+      <ul>
+        <li><code>$cancel</code>, <code>$doctor</code>, <code>$help</code>, <code>$note</code>, <code>$trace</code></li>
+        <li><code>$skill</code></li>
+        <li><code>$hud</code>, <code>$rcs-setup</code>, <code>$configure-notifications</code></li>
+        <li><code>$ralph-init</code> (legacy compatibility helper)</li>
+        <li><code>$worker</code> (internal) - team worker protocol / lifecycle</li>
+      </ul>
+
+      <h2>Usage Examples</h2>
+      <pre><code>$brief "farming simulator for mobile-first Roblox sessions"
+$brief:motivation "rank the retention drivers"
+$blueprint:psych "design from player desire backward"
+$blueprint:loop "define session, daily, weekly, and comeback loops"
+$forge:progression "write the progression ladder"
+$forge:community "write the social retention systems"
+$code-review "review current branch"
+$doctor</code></pre>
+
+      <h2>Visual Verdict Guidance</h2>
+      <ul>
+        <li>Use for visual tasks with reference image(s) + generated screenshot.</li>
+        <li>Pass threshold target: <strong>90+</strong>.</li>
+        <li>Use pixel diff / pixelmatch overlays only as secondary debugging support.</li>
+      </ul>
+    </main>
+
+    <footer>
+      <div class="container"><span class="muted">Browse available skills with <code>/skills</code> in Codex CLI.</span></div>
+    </footer>
+  </body>
+</html>

package/docs/style.css

@@ -0,0 +1,179 @@
+* {
+  box-sizing: border-box;
+}
+
+:root {
+  --bg: #0d1117;
+  --panel: #161b22;
+  --text: #c9d1d9;
+  --muted: #8b949e;
+  --link: #58a6ff;
+  --accent: #238636;
+  --border: #30363d;
+}
+
+html,
+body {
+  margin: 0;
+  padding: 0;
+  background: var(--bg);
+  color: var(--text);
+  font-family: Inter, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, Helvetica, Arial, sans-serif;
+  line-height: 1.6;
+}
+
+a {
+  color: var(--link);
+  text-decoration: none;
+}
+
+a:hover {
+  text-decoration: underline;
+}
+
+header {
+  border-bottom: 1px solid var(--border);
+  background: rgba(13, 17, 23, 0.95);
+  position: sticky;
+  top: 0;
+  backdrop-filter: blur(6px);
+}
+
+.container {
+  width: min(1040px, 92vw);
+  margin: 0 auto;
+}
+
+.nav {
+  display: flex;
+  flex-wrap: wrap;
+  align-items: center;
+  gap: 1rem;
+  padding: 0.9rem 0;
+}
+
+.brand {
+  font-weight: 700;
+  color: var(--text);
+  margin-right: auto;
+}
+
+.nav-links {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 0.8rem;
+}
+
+.nav-links a {
+  padding: 0.35rem 0.6rem;
+  border-radius: 6px;
+}
+
+.nav-links a.active,
+.nav-links a:hover {
+  background: var(--panel);
+}
+
+main {
+  padding: 2rem 0 3rem;
+}
+
+.hero {
+  padding: 2rem;
+  border: 1px solid var(--border);
+  border-radius: 12px;
+  background: linear-gradient(180deg, rgba(35, 134, 54, 0.12), rgba(22, 27, 34, 0.9));
+}
+
+h1,
+h2,
+h3 {
+  line-height: 1.25;
+  margin-top: 0;
+}
+
+h1 {
+  font-size: clamp(1.8rem, 3.8vw, 2.8rem);
+}
+
+h2 {
+  margin-top: 2rem;
+  font-size: 1.5rem;
+}
+
+.card-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(240px, 1fr));
+  gap: 1rem;
+  margin-top: 1rem;
+}
+
+.card {
+  background: var(--panel);
+  border: 1px solid var(--border);
+  border-radius: 10px;
+  padding: 1rem;
+}
+
+.badge {
+  display: inline-block;
+  background: rgba(35, 134, 54, 0.2);
+  color: #7ee787;
+  border: 1px solid rgba(35, 134, 54, 0.5);
+  border-radius: 999px;
+  padding: 0.1rem 0.55rem;
+  font-size: 0.8rem;
+}
+
+code,
+pre {
+  font-family: ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, "Liberation Mono", "Courier New", monospace;
+}
+
+pre {
+  margin: 1rem 0;
+  padding: 1rem;
+  border-radius: 8px;
+  border: 1px solid var(--border);
+  background: #010409;
+  overflow-x: auto;
+}
+
+table {
+  width: 100%;
+  border-collapse: collapse;
+  margin-top: 1rem;
+}
+
+th,
+td {
+  text-align: left;
+  border: 1px solid var(--border);
+  padding: 0.65rem;
+  vertical-align: top;
+}
+
+th {
+  background: #161b22;
+}
+
+.muted {
+  color: var(--muted);
+}
+
+footer {
+  border-top: 1px solid var(--border);
+  padding: 1.2rem 0 2rem;
+  color: var(--muted);
+}
+
+@media (max-width: 700px) {
+  .brand {
+    width: 100%;
+    margin-right: 0;
+  }
+
+  .hero {
+    padding: 1.3rem;
+  }
+}