Spendict
Summary
Your AI ad agent writes a hundred creatives an hour — and without a filter, your spend follows all of them into the test budget before you know which three were worth running. Spendict exists to catch the other ninety-seven before a cent moves.
Spendict issues a deterministic run, fix_first, or kill verdict on every ad creative or campaign structure you feed it, using four discrete tools: creative scoring, campaign structure auditing, live performance diagnosis, and targeting strategy validation. The verdict logic traces back to performance marketers with paid-social backgrounds who calibrated the model against real ads — not synthetic benchmarks. You wire it in via MCP, CLI, npm skill, or REST, and it slots into agents already running in Claude Code, Cursor, Codex, or Gemini. The ceiling appears when your workflow needs verdicts that adapt across iterations or chain decisions across tools autonomously — Spendict returns a single verdict per call and nothing more.
Bottom line: Pick this when your AI agent is generating ad volume faster than your team can manually filter it; look elsewhere when you need the tool to act on its own verdicts, chase fixes, or run multi-step remediation without you driving each call.
Pricing Plans
Usage-Based- Free Tier
- 100 verdicts
Free
100 free verdicts
- 100 verdicts
- MCP/CLI/Skill/REST access
Paid
Usage-based after free tier
- Penny per call
- Same verdicts and quota
View full pricing on spendict.com →
Pricing may have changed since last verified. Check the official site for current plans.
Community Performance Report Card
No community ratings yet. Be the first to rate this tool!
Community Benchmarks Community
Sign in to submit a benchmarkNo community benchmarks yet. Be the first to share a real-world data point.
Pros
Sign in to edit- Four discrete, purpose-built tools covering creative quality, campaign structure, live performance diagnosis, and targeting strategy — so your agent gets a verdict scoped to the actual decision at hand rather than a generic quality score that conflates unrelated failure modes.
- Deterministic run/fix_first/kill output per call, which means downstream agent logic can branch on a string value instead of parsing a confidence interval or summarizing a freeform critique.
- Named failure mode attached to every creative verdict, so when an ad scores kill, the agent — or the human reviewing the queue — knows exactly which dimension failed rather than spending time reverse-engineering the number.
- Four integration surfaces (Skill, MCP, CLI, REST) all sharing the same quota and verdict format, so you connect once in whatever agent framework you already run and avoid re-implementing the client if you migrate environments.
- Per-call pricing at approximately a penny after the free quota, which means the cost of filtering a bad ad before launch is structurally lower than the minimum test budget on any major platform — removing the usual argument for skipping pre-launch review.
Cons
Sign in to edit- Spendict returns one verdict per call and holds no state between calls — so if your workflow requires iterative revision loops where the tool re-evaluates a fix_first creative after edits and tracks improvement, you build and maintain that loop yourself on top of the API.
- The tool does not execute any action on verdict — it cannot pause a campaign, reject a creative in your CMS, or trigger a downstream workflow on its own. Teams expecting the verdict to do anything other than return a string will wire every consequent action manually, which adds integration surface that has to be maintained.
- There is no self-hosted deployment option. Workflows in regulated industries or organizations with strict data-residency requirements that cannot send creative or campaign data to a third-party API will hit this wall immediately and need to evaluate a different architecture entirely.
Community Reviews
Sign in to write a reviewNo reviews yet. Be the first to share your experience.
About
- Platforms
- Web API, CLI, MCP, Skill
- API Available
- Yes
- Self-Hosted
- No
- Last Updated
- 2026-07-13T00:34:47.472Z
Best For
Who it's for
- Teams running AI ad-generation agents
- Performance marketers seeking pre-spend filters
- Agencies scaling high-volume creative testing
What it does well
- Gate every ad creative in an AI agent before launch
- Audit campaign structure and budget allocation pre-launch
- Diagnose live campaigns for fatigue versus structural issues
- Generate platform-validated targeting strategies
Integrations
Discussion Community
Sign in to commentNo discussion yet. Sign in to start the conversation.
Compare Spendict
Spotted incorrect or missing data? Join our community of contributors.
Sign Up to ContributeCommunity Notes & Tips Community
Sign in to contributeBe the first to contribute. General notes, observations, gotchas, and tips from people who use this tool day-to-day.
Frequently Asked Questions
- Is Spendict free?
- Spendict has a permanent free tier alongside paid upgrades. You can keep using a baseline version indefinitely without paying.
- Is Spendict open source?
- No — Spendict is a closed-source tool. Source code is not publicly available.
- Does Spendict have an API?
- Yes. Spendict exposes a developer API. See the official documentation at https://spendict.com for details.
- What platforms does Spendict support?
- Spendict is available on: Web API, CLI, MCP, Skill.
Hours Saved & ROI Stories Community
Sign in to contributeBe the first to contribute. Concrete time/cost savings, with context. e.g. "Cut my code review backlog from 4h to 45m per week."
Curated lists that include this category
AI-generated ad pipelines produce volume that no human review queue can match — and most teams discover this the hard way, after test spend reveals what a pre-launch filter would have caught for a fraction of the cost. Spendict addresses that gap with four API-exposed tools: assess_ad_creative scores hook, angle, clarity, audience fit, platform fit, CTA, and compliance, then names a single predicted failure mode; audit_campaign_structure checks budget allocation, audience setup, bid strategy, and measurement against platform rulesets before fragmentation happens; analyze_campaign_performance separates creative fatigue from structural problems using live metrics; and strategize_targeting returns a structure-validated targeting plan your agent can build from directly.
The differentiating claim is determinism. Each call returns one of a fixed set of verdicts — run, fix_first, kill, or their campaign-layer equivalents — rather than a probability score or a paragraph of suggestions. The vendor states the grading methodology was developed with and calibrated by performance marketers with 10+ years in paid social, and the landing page exposes scored examples with attached failure modes rather than hiding the reasoning behind the number.
Spendict is designed for one specific position in an agent workflow: the gate before spend is committed, not the system that runs the campaign. It does not execute changes, manage creative iteration, or persist context across calls. Teams that need the tool to chase a fix_first verdict through revision rounds, or to autonomously re-audit after changes, will have to build that loop themselves — Spendict returns a verdict and waits for the next call.
Integration follows four paths the docs describe as equivalent in quota and verdict output: an npm-installable Skill that works in Claude Code, Cursor, Codex, and Gemini via a single SKILL.md file; an MCP one-click connector requiring no API key management; a CLI installed via npm; and a plain REST endpoint using bearer key authentication. The vendor states 100 verdicts are included at no cost, with additional calls priced at approximately a penny each.