Agatha All Along and the MCU's Last Competent Show

Agatha All Along aired on Disney+ across nine episodes in September and October 2024. Jac Schaeffer created and ran it, continuing the thread she started on WandaVision. Kathryn Hahn returned as Agatha Harkness, the sitcom-era witch from Westview. The show’s premise was narrow: Agatha, stripped of her powers at the end of WandaVision, assembles a coven to walk the Witches’ Road, a mythical route whose completion is supposed to restore what she has lost.

I am going to say this directly: Agatha All Along is the best thing the MCU has put on television in three years, and it is also, on current evidence, probably the last genuinely competent piece of franchise television Marvel will release this decade. Nine months on, with the subsequent Marvel output continuing to disappoint, this one looks sharper.

Why this show worked when almost nothing else Marvel has done has

The specific reason Agatha All Along works is the same specific reason WandaVision worked: Jac Schaeffer has a clear sense of what kind of show she is making, and the show never drifts from that clarity. This sounds like a low bar. It is a low bar. The Marvel television slate of the last four years has repeatedly cleared it by inches or failed it entirely. Moon Knight did not know what it was. She-Hulk did not know what it was. Secret Invasion did not know what it was. Echo knew for about four episodes and then stopped knowing.

Schaeffer knows. Agatha All Along is a road show, built around a coven, drawing on the specific folk-horror iconography of the Witches’ Road as a narrative device. Each episode has a specific internal logic (an obstacle, a ritual test, a character revelation) that corresponds to a specific tarot-arcana register. The show is willing to commit to witchcraft as its genre, rather than hedging the witchcraft with continuous MCU-universe cross-referencing.

This is the other specific thing it does right: it does not drown in cameos. The show is not a vehicle for setting up five other Marvel projects. It stays in its lane.

Kathryn Hahn, earning the title card

Kathryn Hahn was the best thing in WandaVision. She is also the best thing in Agatha All Along, and the best thing by a wider margin because this show has been built specifically as a vehicle for her. Hahn plays Agatha with a specific theatrical-camp register that modern prestige TV has mostly abandoned because prestige TV has become allergic to scenery-chewing. Hahn does not care. She chews the scenery. She has always chewed the scenery. The chewing is the character.

What Hahn adds in this show that she did not quite have access to in WandaVision is a specific register of wounded vanity. Agatha has spent three centuries accumulating power, and the show’s premise strips her of it. Hahn plays the stripped version as a woman whose whole self-conception has been tied to her specific magical seniority, and who is now forced to operate among younger, less experienced witches as a kind of deposed courtier. The performance is comic and dramatic at the same time.

The scene that locates Hahn’s whole work here is in the fifth episode. Agatha has just been tricked, specifically, by one of the younger coven members. She does not react immediately. The camera holds on her for about eight seconds of her face doing a specific internal accounting. Hahn plays it as a woman who is cataloguing the humiliation, filing it for later use, and deciding, simultaneously, not to give the other witch the satisfaction of a visible response. This is character acting at a level the MCU has rarely accessed.

The coven

The supporting coven is where the show spends its character budget. Aubrey Plaza’s Rio Vidal, Patti LuPone’s Lilia Calderu, Sasheer Zamata’s Jennifer Kale, Ali Ahn’s Alice Wu-Gulliver, and Joe Locke as the unnamed Teen (later revealed, but I will not spoil here) are each given, across the nine episodes, at least one dedicated character episode.

Patti LuPone is the performance I want to mark specifically. LuPone, whose stage work has always been at a higher register than most of her screen work has required, plays Lilia Calderu as a specific kind of witch: ancient, tarot-trained, loose in time. The character, in the sixth episode, gets an extended sequence of her experiencing her own life non-linearly. LuPone plays the scene at the specific physical register of a theatrical veteran who has been handed a good monologue. It is the best single piece of LuPone screen acting in years, and Schaeffer’s episode structure gives it the space it needs.

Aubrey Plaza, as Rio Vidal, is the show’s other structural performance. Rio is the character we are supposed to read one way for most of the show and then read differently after a specific mid-season revelation. Plaza plays the doubling with a specific watchful affect that has become her best acting register. She is not trying to do anything. The not-trying is the work.

Folk horror, specifically

What the show is actually drawing on, at the level of visual language and sound design, is the British-American folk-horror tradition. The Wicker Man (the real one, not the remake). Midsommar. Häxan. Witchfinder General. The show’s second episode, which stages the coven’s first ritual, is composed with a specific attention to the iconography of folk-horror cinema: the circle of candles, the specific hand positions, the specific colour palette of the staging.

I am not saying the show is as good as the films it is drawing on. It is not. It is a Disney property, and it has to operate within specific aesthetic and narrative constraints. But it draws on the tradition with a specific literacy that distinguishes it from the MCU’s previous attempts at genre specificity. Compare it to Moon Knight, which gestured at Egyptian mythology and horror and never settled on either. Agatha All Along commits.

The music, which is doing structural work

The show’s theme song, “The Ballad of the Witches’ Road,” written by Kristen Anderson-Lopez and Robert Lopez (the WandaVision theme team), is performed by the coven at multiple moments across the season, with different arrangements reflecting specific emotional states. This is a formal conceit that should not work and does. The repeated song functions as the show’s structural spine, and the specific moment in the finale when the song returns in a different register is the emotional pivot the whole season has been building to.

This is musical-dramatic work at a specific level. The song is not a marketing object. It is a scene.

Where the show stumbles

The show’s central structural weakness is its relationship to continuity. Agatha’s backstory is revealed across the season in ways that sometimes conflict with what was established in WandaVision. Schaeffer has resolved most of these conflicts cleanly, but some viewers who came to the show expecting strict continuity were frustrated. I am not one of those viewers. I think Schaeffer’s looseness with prior franchise continuity is actually one of the show’s assets. But it is a genuine point of contention and worth naming.

The other weakness is the last episode. The finale, like most MCU finales, has to deliver a specific plot-level resolution that is partly being driven by larger franchise concerns. The resolution is handled more gracefully than most MCU finales, but it still feels slightly constrained.

What the show leaves

Agatha All Along is the show that demonstrated that Marvel television can still work when the people running it are allowed to commit to a specific genre register and when the casting prioritises actors who can carry the weight. Schaeffer has said she does not know yet whether she will continue in the MCU. Given what has happened to the Marvel TV slate subsequent to her work, I would advise her strongly against it. She has other skills. The franchise will keep misusing them.

Watch the season in two sittings. Pay attention to the song’s repetitions. The show is at its best when it trusts its cast, its source tradition, and its specific theatrical instincts. In a franchise that has mostly forgotten how to do any of those three, it is the one remaining argument for the form.