How to Batch Create and Schedule Instagram Reels with Claude Code and Wonda

The hard part of Instagram is not generating one Reel.

It is keeping a queue of good Reels ready to publish without turning content work into a second full-time job.

If you want a practical answer: Claude Code is the operator, Wonda is the execution layer, and the repeatable unit is a reviewed Reel queue. You tell Claude Code what batch you want, it runs the Wonda commands, compares outputs, drafts the publishing queue, and prepares the keepers for scheduling. You keep the final judgment on what actually goes live.

That split is the whole point. You do not need another social dashboard. You need an agent that can operate a real toolchain from the same terminal where you already work.

If you have already read You Don't Need to Learn the CLI: Let Claude Code Run Wonda for You, this is the Instagram-specific version of that idea. The workflow here is narrower and more practical: batch a week of Instagram Reels, review them, and schedule them safely.

Key Takeaways

• Claude Code is best used here as an operator for planning, generation, review, and scheduling.
• Wonda gives Claude Code the actual Instagram commands: generate, edit, publish, and account inspection.
• The only Instagram constraints that really matter in the workflow are the content_publishing_limit check and the 24-hour container window.
• The safest setup is still human-reviewed on every publish action.

What This Article Is Actually About

This article is specifically about batching Instagram Reels with Claude Code and Wonda.

It is not a general Instagram growth guide, and it is not a dashboard roundup. The use case here is narrower: you already work in the terminal, you want Claude Code to operate the workflow, and you want Wonda to handle generation, edits, publishing, and analytics.

If you want the broader control-layer argument, read You Don't Need to Learn the CLI: Let Claude Code Run Wonda for You. If you want the lower-level publish fundamentals, read How to Automate Instagram Posting from the Terminal with AI Agents.

Why This Workflow Fits Claude Code Better Than a Dashboard

Most Instagram tools are built around clicking.

You plan in one place, upload in another, edit somewhere else, schedule in another tab, and keep a spreadsheet open the whole time because none of the interfaces actually compose well.

Claude Code plus Wonda compresses that loop:

1. define the batch in plain English
2. generate the assets
3. inspect the results
4. caption and prepare the keepers
5. schedule or publish them

That is a much better fit for the kind of work Instagram actually requires when you are doing it seriously. The job is not "open composer, type caption." The job is "keep a queue of usable creative moving without losing taste."

What Claude Code and Wonda Each Handle

Claude Code is good at:

• turning a batch brief into a concrete plan
• writing and tightening prompts
• deciding which commands to run in what order
• inspecting outputs and comparing variants
• drafting captions, alt text, and schedule manifests

Wonda is good at giving Claude Code a real execution surface:

• wonda brand
• wonda accounts instagram
• wonda scrape social
• wonda generate video
• wonda generate image
• wonda edit video
• wonda publish instagram
• wonda publish instagram-carousel
• wonda analytics instagram

That distinction matters. You do not want Claude Code inventing an Instagram workflow. You want it operating a real one.

Setup

Install Wonda:

curl -fsSL https://wonda.sh/install.sh | bash
wonda auth login

Make sure Claude Code can see the Instagram command surface:

wonda accounts instagram
wonda publish instagram --help

If you are running this inside Claude Code already, that is enough. Wonda is the execution layer. Claude Code is the control layer.

Step 1: Give Claude Code a Real Batch Brief

Do not start with "make me some Instagram content."

Start with a brief that gives Claude Code enough structure to make good decisions.

Example:

Help me batch 5 Instagram Reels for next week for my coffee brand.
Target: founders and operators.
Tone: calm, minimal, not hypey.
Format: 8-12 second vertical Reels.
Goal: save-worthy morning routine content.

Use Wonda to:
1. load brand context
2. inspect one or two competitor accounts
3. draft 5 Reel concepts
4. generate the clips
5. add animated captions
6. prepare captions and alt text
7. schedule the best 3 and leave the weaker 2 as drafts for revision

That prompt gives Claude Code three things it actually needs:

• the audience
• the creative direction
• the operational outcome

Without those, you just get fast generic output.

Step 2: Let Claude Code Build the Batch

The first commands should usually be context commands, not generation commands:

wonda brand
wonda scrape social --handle @competitor --platform instagram --wait

From there, Claude Code can move into generation. For a text-to-video Reel, the pattern is straightforward:

VID_JOB=$(wonda generate video \
  --model sora2 \
  --prompt "slow vertical push-in on espresso pouring, soft morning light, shallow depth, minimal kitchen setting" \
  --aspect-ratio 9:16 \
  --duration 8 \
  --wait \
  --quiet)

VID_MEDIA=$(wonda jobs get inference "$VID_JOB" --jq '.outputs[0].media.mediaId')

If the Reel depends on a reference image with a visible face, Claude Code should route to kling_3_pro instead:

REF=$(wonda media upload ./reference.png --quiet)

VID_JOB=$(wonda generate video \
  --model kling_3_pro \
  --attach "$REF" \
  --prompt "subtle breathing motion, natural blinking, slow camera push-in" \
  --aspect-ratio 9:16 \
  --duration 5 \
  --wait \
  --quiet)

VID_MEDIA=$(wonda jobs get inference "$VID_JOB" --jq '.outputs[0].media.mediaId')

That is the part Claude Code is good at: taking a mixed batch brief and routing each asset correctly instead of forcing one model across every Reel.

Step 3: Review the Batch Before It Becomes a Schedule

This is the step most "AI social automation" content gets wrong.

The value is not that Claude Code can generate five Reels. The value is that it can inspect five Reels and help you pick the two or three that are actually worth publishing.

For Instagram, the fastest edit pass is usually animated captions:

CAP_JOB=$(wonda edit video \
  --operation animatedCaptions \
  --media "$VID_MEDIA" \
  --params '{"fontFamily":"TikTok Sans SemiCondensed","position":"bottom-center","sizePercent":80,"strokeWidth":2.5,"fontSizeScale":0.8,"highlightColor":"rgb(252, 61, 61)"}' \
  --wait \
  --quiet)

FINAL_MEDIA=$(wonda jobs get editor "$CAP_JOB" --jq '.outputs[0].mediaId')

Then ask Claude Code to evaluate the outputs in plain language:

Review these 5 generated Reels.
Tell me:
- which 3 have the strongest first two seconds
- which ones feel too generic
- which captions need rewriting
- which one should not be scheduled yet

That keeps the workflow aligned with what we actually want Wonda to do: make an agent more useful as an operator, not turn the whole thing into blind autopilot.

Step 4: Turn the Winners Into a Real Schedule

This is the one place where Instagram-specific mechanics matter.

You do not need a giant section on the entire Instagram ecosystem. You need two facts that actually change how you schedule:

• Meta says Instagram accounts are limited to 100 API-published posts within a 24-hour moving period, and carousels count as a single post. See Meta's content publishing docs.
• Meta also marks a container as EXPIRED if it is not published within 24 hours. See the container status reference.

That leads to a simpler rule than most people use: do not try to "schedule containers" far in advance. Keep approved Wonda media ready, keep the schedule in a manifest, and let Claude Code trigger wonda publish instagram close to the actual slot.

A simple queue file is enough:

publish_at,media_id,caption,alt_text
2026-04-28T09:00:00+02:00,med_01,Tuesday reset. What's your first move of the day?,Slow vertical push on espresso pouring in warm morning light
2026-04-30T09:00:00+02:00,med_02,Three quiet ways to make your morning feel less rushed.,Close shot of coffee tools on a minimal kitchen counter
2026-05-02T10:00:00+02:00,med_03,The five-minute ritual we would actually repeat.,Hands preparing coffee in soft natural window light

Claude Code can generate that file after the review pass, rewrite weak captions, and drop any Reel that does not deserve a slot.

Before each publish window, the pre-flight check should look like this:

USAGE=$(curl -s \
  "https://graph.facebook.com/v19.0/${IG_USER_ID}/content_publishing_limit?access_token=${ACCESS_TOKEN}" \
  | jq -r '.data[0].quota_usage')

if [ "$USAGE" -ge 90 ]; then
  echo "Near publish cap (used: $USAGE). Delay the rest of the batch."
fi

Then the scheduled command stays simple and immediate:

wonda publish instagram \
  --media "$FINAL_MEDIA" \
  --account "$IG_ACCOUNT" \
  --caption "Tuesday reset. What's your first move of the day?" \
  --alt-text "Slow vertical push on espresso pouring in warm morning light" \
  --share-to-feed

The scheduling move is not that Wonda magically turns one command into a social calendar. The move is that Claude Code can maintain the queue and hand each approved row to a scheduler at the right time. That can be a cron job, a CI schedule, or Wonda's scheduling surface if that is where you want the final handoff.

That is a much truer version of "schedule with Claude Code." It means the agent owns the batch state, not that you leave long-lived Instagram containers hanging around.

A Good Weekly Operating Loop

For most teams, the sane version of this is a weekly loop, not continuous posting.

Example:

1. Monday: ask Claude Code to plan 5-7 Reels from one brand angle.
2. Tuesday: generate the batch with Wonda.
3. Wednesday: review the outputs, reject the weak ones, tighten captions, and write the queue file.
4. Thursday: let the scheduler trigger wonda publish instagram for the approved slots.
5. Friday: pull performance with wonda analytics instagram and feed the strongest hooks back into next week's batch.

The command for the review side is lightweight:

wonda analytics instagram

Then ask Claude Code:

Look at last week's Instagram performance and tell me:
- which hooks got reused attention
- which captions underperformed
- what pattern should carry into next week's batch

That is the loop we actually care about. Not "post more." Learn faster, keep quality higher, and make batching less manual.

What Still Needs Human Judgment

Taste

Claude Code can sort outputs. It still cannot fully replace taste. You still need someone to decide whether a Reel feels on-brand or merely acceptable.

Risk

Do not let an agent publish controversial or reputation-sensitive content unattended.

Voice

Caption drafts are useful. Final voice still benefits from a human pass, especially on Instagram where flat generic language dies quickly.

Volume

If the workflow starts pushing you toward more posting just because it is easier, tighten the loop. Better batches beat bigger batches.

FAQ

Is this article about Claude Code or about the Wonda CLI?

Both, but the split is clear: Claude Code is the operator, Wonda is the execution surface. The useful part is not memorizing commands. The useful part is that Claude Code can run them for you from a real skill-aware CLI.

Do I need to write bash scripts for this to work?

No. The bash examples show the command shape underneath the workflow. In practice, the point is that you can describe the batch in natural language and let Claude Code translate it into commands. If you do want a reusable script later, Claude Code can write that too.

Can Claude Code schedule the content as well as generate it?

Yes, but the reliable unit to schedule is the publish command, not a long-lived Instagram container. The useful version is: plan the batch, generate the assets, review them, write the queue, then let Claude Code hand approved rows to a scheduler or to Wonda's scheduling surface near publish time.

What should I automate first?

Start with one weekly Reel batch, not a full social calendar. One narrow loop is enough to prove whether the prompts, model routing, caption style, and schedule rhythm are actually working.

The Bottom Line

If the workflow is right, Claude Code should make Instagram batching feel smaller, not bigger.

You tell it what kind of Reel batch you want. It runs Wonda to generate the assets, inspect the outputs, prepare the captions, write the publishing queue, and hand the keepers to the right scheduling surface. You keep the parts that actually need judgment: taste, voice, and final approval.

That is much closer to what we actually do than a giant generic "Instagram batching in 2026" article.

If you want the product surface this workflow runs on, start with Wonda for Instagram, Claude Code for Instagram, and Wonda's scheduling surface.