diff --git a/plots/acf-pacf/implementations/julia/makie.jl b/plots/acf-pacf/implementations/julia/makie.jl
new file mode 100644
index 0000000000..db556695bc
--- /dev/null
+++ b/plots/acf-pacf/implementations/julia/makie.jl
@@ -0,0 +1,170 @@
+# anyplot.ai
+# acf-pacf: Autocorrelation and Partial Autocorrelation (ACF/PACF) Plot
+# Library: makie 0.22.10 | Julia 1.11.9
+# Quality: 86/100 | Created: 2026-06-10
+
+using CairoMakie
+using Colors
+using Random
+using Statistics
+
+Random.seed!(42)
+
+# Theme tokens
+const THEME     = get(ENV, "ANYPLOT_THEME", "light")
+const PAGE_BG   = THEME == "light" ? colorant"#FAF8F1" : colorant"#1A1A17"
+const INK       = THEME == "light" ? colorant"#1A1A17" : colorant"#F0EFE8"
+const INK_SOFT  = THEME == "light" ? colorant"#4A4A44" : colorant"#B8B7B0"
+const IMPRINT_PALETTE = [
+    colorant"#009E73", colorant"#C475FD", colorant"#4467A3", colorant"#BD8233",
+    colorant"#AE3030", colorant"#2ABCCD", colorant"#954477", colorant"#99B314",
+]
+
+# Data: AR(2) time series — monthly economic indicator
+# x[t] = 0.7*x[t-1] + 0.2*x[t-2] + noise  (φ₁=0.7, φ₂=0.2)
+# PACF cuts off at lag 2; ACF decays as a mixture of exponentials
+n_obs = 300
+series = zeros(n_obs)
+series[1] = randn()
+series[2] = 0.7 * series[1] + randn()
+for t in 3:n_obs
+    series[t] = 0.7 * series[t-1] + 0.2 * series[t-2] + randn()
+end
+
+# ACF (lags 0 through max_lag)
+max_lag = 35
+s_mean = mean(series)
+s_var  = sum((v - s_mean)^2 for v in series) / n_obs
+acf_vals = [sum((series[i] - s_mean) * (series[i-k] - s_mean) for i in (k+1):n_obs) / (n_obs * s_var)
+            for k in 0:max_lag]
+
+# PACF via Levinson-Durbin recursion
+phi       = zeros(max_lag, max_lag)
+phi[1, 1] = acf_vals[2]
+pacf_vals = zeros(max_lag)
+pacf_vals[1] = phi[1, 1]
+for k in 2:max_lag
+    numer = acf_vals[k + 1]
+    denom = 1.0
+    for j in 1:(k - 1)
+        numer -= phi[k-1, j] * acf_vals[k - j + 1]
+        denom -= phi[k-1, j] * acf_vals[j + 1]
+    end
+    phi[k, k] = numer / denom
+    for j in 1:(k - 1)
+        phi[k, j] = phi[k-1, j] - phi[k, k] * phi[k-1, k-j]
+    end
+    pacf_vals[k] = phi[k, k]
+end
+
+conf_bound = 1.96 / sqrt(n_obs)
+acf_lags   = Float64.(0:max_lag)
+
+# Build linesegments! data for ACF: interleaved [x_base, x_tip, ...] pairs per segment
+acf_seg_x = vcat([[lag, lag] for lag in acf_lags]...)
+acf_seg_y = vcat([[0.0, v]   for v   in acf_vals]...)
+
+# PACF: split lags into significant (|r| > conf_bound) vs. within-bound for visual hierarchy
+pacf_sig_idx = findall(v -> abs(v) > conf_bound, pacf_vals)
+pacf_ns_idx  = findall(v -> abs(v) <= conf_bound, pacf_vals)
+
+# Muted variant for non-significant PACF lags (same hue, lower opacity)
+GREEN_FULL  = IMPRINT_PALETTE[1]
+GREEN_MUTED = RGBAf(GREEN_FULL.r, GREEN_FULL.g, GREEN_FULL.b, 0.38f0)
+
+function make_segs(idx, vals)
+    xs = vcat([[Float64(i), Float64(i)] for i in idx]...)
+    ys = vcat([[0.0, vals[i]] for i in idx]...)
+    xs, ys
+end
+
+# Figure
+fig = Figure(
+    size            = (1600, 900),
+    fontsize        = 14,
+    backgroundcolor = PAGE_BG,
+)
+
+# ACF axis (top)
+ax_acf = Axis(
+    fig[1, 1];
+    title             = "AR(2) Process · acf-pacf · julia · makie · anyplot.ai",
+    titlesize         = 20,
+    titlecolor        = INK,
+    ylabel            = "ACF",
+    ylabelsize        = 14,
+    ylabelcolor       = INK,
+    xticklabelsize    = 12,
+    yticklabelsize    = 12,
+    xticklabelcolor   = INK_SOFT,
+    yticklabelcolor   = INK_SOFT,
+    xtickcolor        = INK_SOFT,
+    ytickcolor        = INK_SOFT,
+    backgroundcolor   = PAGE_BG,
+    topspinevisible   = false,
+    rightspinevisible = false,
+    leftspinecolor    = INK_SOFT,
+    bottomspinecolor  = INK_SOFT,
+    xgridvisible      = false,
+    ygridvisible      = true,
+    ygridcolor        = RGBAf(INK.r, INK.g, INK.b, 0.12f0),
+)
+
+# PACF axis (bottom)
+ax_pacf = Axis(
+    fig[2, 1];
+    xlabel            = "Lag",
+    ylabel            = "PACF",
+    xlabelsize        = 14,
+    ylabelsize        = 14,
+    xlabelcolor       = INK,
+    ylabelcolor       = INK,
+    xticklabelsize    = 12,
+    yticklabelsize    = 12,
+    xticklabelcolor   = INK_SOFT,
+    yticklabelcolor   = INK_SOFT,
+    xtickcolor        = INK_SOFT,
+    ytickcolor        = INK_SOFT,
+    backgroundcolor   = PAGE_BG,
+    topspinevisible   = false,
+    rightspinevisible = false,
+    leftspinecolor    = INK_SOFT,
+    bottomspinecolor  = INK_SOFT,
+    xgridvisible      = false,
+    ygridvisible      = true,
+    ygridcolor        = RGBAf(INK.r, INK.g, INK.b, 0.12f0),
+)
+
+linkxaxes!(ax_acf, ax_pacf)
+ax_acf.xticklabelsvisible = false
+ax_acf.xlabelvisible      = false
+
+# ACF stems, tips, and reference lines
+linesegments!(ax_acf, acf_seg_x, acf_seg_y;
+    color = GREEN_FULL, linewidth = 2.0)
+scatter!(ax_acf, acf_lags, acf_vals;
+    color = GREEN_FULL, markersize = 8, strokewidth = 0)
+hlines!(ax_acf, [0.0]; color = INK_SOFT, linewidth = 1.0)
+hlines!(ax_acf, [conf_bound, -conf_bound];
+    color = INK_SOFT, linestyle = :dash, linewidth = 1.5)
+
+# PACF: within-bound lags (muted) drawn first, then significant lags on top
+if !isempty(pacf_ns_idx)
+    ns_sx, ns_sy = make_segs(pacf_ns_idx, pacf_vals)
+    linesegments!(ax_pacf, ns_sx, ns_sy; color = GREEN_MUTED, linewidth = 1.5)
+    scatter!(ax_pacf, Float64.(pacf_ns_idx), pacf_vals[pacf_ns_idx];
+        color = GREEN_MUTED, markersize = 7, strokewidth = 0)
+end
+if !isempty(pacf_sig_idx)
+    sig_sx, sig_sy = make_segs(pacf_sig_idx, pacf_vals)
+    linesegments!(ax_pacf, sig_sx, sig_sy; color = GREEN_FULL, linewidth = 2.5)
+    scatter!(ax_pacf, Float64.(pacf_sig_idx), pacf_vals[pacf_sig_idx];
+        color = GREEN_FULL, markersize = 9, strokewidth = 0)
+end
+hlines!(ax_pacf, [0.0]; color = INK_SOFT, linewidth = 1.0)
+hlines!(ax_pacf, [conf_bound, -conf_bound];
+    color = INK_SOFT, linestyle = :dash, linewidth = 1.5)
+
+rowgap!(fig.layout, 1, 20)
+
+save("plot-$(THEME).png", fig; px_per_unit = 2)
diff --git a/plots/acf-pacf/metadata/julia/makie.yaml b/plots/acf-pacf/metadata/julia/makie.yaml
new file mode 100644
index 0000000000..cded70e7e9
--- /dev/null
+++ b/plots/acf-pacf/metadata/julia/makie.yaml
@@ -0,0 +1,249 @@
+library: makie
+language: julia
+specification_id: acf-pacf
+created: '2026-06-10T02:02:27Z'
+updated: '2026-06-10T02:25:19Z'
+generated_by: claude-sonnet
+workflow_run: 27247698933
+issue: 4663
+language_version: 1.11.9
+library_version: 0.22.10
+preview_url_light: https://storage.googleapis.com/anyplot-images/plots/acf-pacf/julia/makie/plot-light.png
+preview_url_dark: https://storage.googleapis.com/anyplot-images/plots/acf-pacf/julia/makie/plot-dark.png
+preview_html_light: null
+preview_html_dark: null
+quality_score: 86
+review:
+  strengths:
+  - 'Excellent spec compliance: all required features present including shared x-axis
+    via linkxaxes!, vertical stems with dot tips, dashed 95% confidence bounds, lag
+    0 in ACF but not PACF'
+  - 'Strong data storytelling: AR(2) process choice beautifully illustrates the textbook
+    ACF/PACF signatures — slow exponential ACF decay contrasted with sharp PACF cutoff
+    at lag 2'
+  - 'Visual hierarchy for PACF: muted alpha (0.38) for non-significant lags vs full
+    #009E73 for significant lags — immediately guides the eye to the identifying AR(2)
+    spikes'
+  - 'Full theme-adaptive chrome: all six Makie color tokens (PAGE_BG, INK, INK_SOFT,
+    ygridcolor) correctly wired — both renders are clean and readable'
+  - Levinson-Durbin recursion implemented from scratch — no statsmodels/scipy dependency
+    needed, keeps implementation self-contained
+  weaknesses:
+  - Helper function make_segs() at line 75 violates the KISS no-functions/classes
+    rule — inline the two-line segment construction directly at each call site
+  - 'Design Excellence ceiling: plot reads as a well-configured professional chart
+    but lacks the typographic weight variation (bold title, lighter tick labels) and
+    deliberate whitespace that would push it toward publication-ready'
+  image_description: |-
+    Light render (plot-light.png):
+      Background: Warm off-white #FAF8F1 — correct, not pure white
+      Chrome: Title "AR(2) Process · acf-pacf · julia · makie · anyplot.ai" in dark ink, clearly readable. "ACF" and "PACF" y-axis labels in dark ink. "Lag" x-axis label (bottom panel only) in dark ink. Tick labels in INK_SOFT (#4A4A44). All text fully readable.
+      Data: Single color #009E73 (Imprint brand green) for ACF stems/dots and significant PACF stems/dots. Non-significant PACF lags rendered in same hue at alpha=0.38 — visually muted but legible. Dashed confidence bounds at ~±0.113. Horizontal zero baseline in INK_SOFT. Top and right spines removed, subtle y-axis grid with alpha=0.12.
+      Legibility verdict: PASS — all text readable at desktop and estimated mobile scale
+
+    Dark render (plot-dark.png):
+      Background: Warm near-black #1A1A17 — correct, not pure black
+      Chrome: Title, axis labels, tick labels all in light colors (INK = #F0EFE8 for primary, INK_SOFT = #B8B7B0 for secondary). No dark-on-dark failures observed — text is clearly legible against the dark surface.
+      Data: #009E73 data stems/dots identical to light render — color identity preserved across themes. Confidence bound dashes and zero baseline appear in theme-appropriate light tones. Y-grid is faint but visible.
+      Legibility verdict: PASS — no dark-on-dark issues, all text readable
+  criteria_checklist:
+    visual_quality:
+      score: 28
+      max: 30
+      items:
+      - id: VQ-01
+        name: Text Legibility
+        score: 7
+        max: 8
+        passed: true
+        comment: All font sizes explicitly set (titlesize=20, labels=14, ticks=12).
+          Both renders readable. Near-perfect proportions.
+      - id: VQ-02
+        name: No Overlap
+        score: 6
+        max: 6
+        passed: true
+        comment: No overlapping elements across 36 lags in two panels
+      - id: VQ-03
+        name: Element Visibility
+        score: 5
+        max: 6
+        passed: true
+        comment: Stems and tips clearly visible. Muted PACF dots (alpha=0.38, markersize=7)
+          are visible but slightly small.
+      - id: VQ-04
+        name: Color Accessibility
+        score: 2
+        max: 2
+        passed: true
+        comment: 'Single hue #009E73 with alpha differentiation — no red-green issue,
+          good contrast both backgrounds'
+      - id: VQ-05
+        name: Layout & Canvas
+        score: 4
+        max: 4
+        passed: true
+        comment: Two panels fill canvas well, rowgap=20 gives breathing room, no content
+          cut off. 3200x1800 confirmed.
+      - id: VQ-06
+        name: Axis Labels & Title
+        score: 2
+        max: 2
+        passed: true
+        comment: ACF/PACF are standard accepted abbreviations, Lag is clear. Descriptive
+          title prefix 'AR(2) Process'.
+      - id: VQ-07
+        name: Palette Compliance
+        score: 2
+        max: 2
+        passed: true
+        comment: 'First series #009E73, backgrounds #FAF8F1/#1A1A17, full theme-adaptive
+          chrome. Data colors identical across themes.'
+    design_excellence:
+      score: 13
+      max: 20
+      items:
+      - id: DE-01
+        name: Aesthetic Sophistication
+        score: 5
+        max: 8
+        passed: true
+        comment: 'Above well-configured default: intentional significant/non-significant
+          visual hierarchy, clean spine removal, professional Imprint palette application.
+          Not yet FiveThirtyEight-level.'
+      - id: DE-02
+        name: Visual Refinement
+        score: 4
+        max: 6
+        passed: true
+        comment: Top/right spines removed, subtle y-grid (alpha=0.12), dashed confidence
+          lines add visual interest. Good but not perfect.
+      - id: DE-03
+        name: Data Storytelling
+        score: 4
+        max: 6
+        passed: true
+        comment: AR(2) process choice is pedagogically intentional — slow ACF decay
+          vs PACF cutoff at lag 2. Visual emphasis on significant lags creates a clear
+          focal point.
+    spec_compliance:
+      score: 15
+      max: 15
+      items:
+      - id: SC-01
+        name: Plot Type
+        score: 5
+        max: 5
+        passed: true
+        comment: Two-panel ACF/PACF stem plot — exactly the specified chart type
+      - id: SC-02
+        name: Required Features
+        score: 4
+        max: 4
+        passed: true
+        comment: Shared x-axis (linkxaxes!), 95% CI dashed lines, lag 0 in ACF only,
+          stems+tips, 35 lags
+      - id: SC-03
+        name: Data Mapping
+        score: 3
+        max: 3
+        passed: true
+        comment: X=Lag, Y=correlation values, all data visible
+      - id: SC-04
+        name: Title & Legend
+        score: 3
+        max: 3
+        passed: true
+        comment: 'Title: ''AR(2) Process · acf-pacf · julia · makie · anyplot.ai''
+          — correct format with descriptive prefix. No legend needed.'
+    data_quality:
+      score: 14
+      max: 15
+      items:
+      - id: DQ-01
+        name: Feature Coverage
+        score: 6
+        max: 6
+        passed: true
+        comment: AR(2) shows both slow ACF decay and PACF cutoff — demonstrates all
+          distinguishing features of the plot type
+      - id: DQ-02
+        name: Realistic Context
+        score: 4
+        max: 5
+        passed: true
+        comment: Monthly economic indicator context in comments, realistic AR(2) parameters
+          (phi1=0.7, phi2=0.2). Good but context only in comments, not axis labels.
+      - id: DQ-03
+        name: Appropriate Scale
+        score: 4
+        max: 4
+        passed: true
+        comment: 300 observations (within 100-500 spec), correlation values in [-1,1],
+          confidence bounds at correct 1.96/sqrt(300)
+    code_quality:
+      score: 9
+      max: 10
+      items:
+      - id: CQ-01
+        name: KISS Structure
+        score: 2
+        max: 3
+        passed: true
+        comment: Defines make_segs() helper function at line 75 — violates no-functions
+          rule. Small and practical but still a function.
+      - id: CQ-02
+        name: Reproducibility
+        score: 2
+        max: 2
+        passed: true
+        comment: Random.seed!(42) at line 11
+      - id: CQ-03
+        name: Clean Imports
+        score: 2
+        max: 2
+        passed: true
+        comment: CairoMakie (plotting), Colors (colorant), Random (seed), Statistics
+          (mean) — all used
+      - id: CQ-04
+        name: Code Elegance
+        score: 2
+        max: 2
+        passed: true
+        comment: Clean Levinson-Durbin recursion, logical separation of significant
+          vs non-significant lags, no fake interactivity
+      - id: CQ-05
+        name: Output & API
+        score: 1
+        max: 1
+        passed: true
+        comment: save("plot-$(THEME).png", fig; px_per_unit = 2) — correct
+    library_mastery:
+      score: 7
+      max: 10
+      items:
+      - id: LM-01
+        name: Idiomatic Usage
+        score: 4
+        max: 5
+        passed: true
+        comment: linkxaxes!, linesegments!, scatter!, hlines!, rowgap!, RGBAf — very
+          good idiomatic Makie usage
+      - id: LM-02
+        name: Distinctive Features
+        score: 3
+        max: 5
+        passed: true
+        comment: linkxaxes! for shared axes, Makie layout system (fig[1,1]/fig[2,1]),
+          rowgap!, RGBAf alpha colors — Makie-specific patterns used well
+  verdict: APPROVED
+impl_tags:
+  dependencies: []
+  techniques:
+  - subplots
+  patterns:
+  - data-generation
+  dataprep:
+  - time-series
+  styling:
+  - alpha-blending