Xcode Build Optimization Agent Skill
Speed up Xcode clean and incremental builds through benchmarks and build-setting optimizations.
Evidence: README explicitly states 'No project files are modified until you explicitly approve changes' and the workflow is split into analysis and fix phases, with the fix phase requiring user approval, supporting least privilege and user confirmation. Data flow transparency: README describes the workflow and artifacts (.build-benchmark/optimization-plan.md), but does not detail how scripts handle project data or whether external uploads occur. Sensitive data handling: not mentioned, but as a build optimization tool, it may access project files, yet no explanation of handling sensitive information. Dependency security: no dependency list or security audit provided, but as an Agent Skill, dependencies may be minimal. External effects: clearly states modifications require approval, and provides a rollback mechanism (re-benchmarking), but no automatic rollback. Source attribution: README clearly identifies author as Antoine van der Lee and provides personal website, but publisher is unverified. Deductions: insufficient data flow transparency, sensitive data handling not addressed, dependency security not assessed, rollback mechanism incomplete.
Evidence: README and skill structure are consistent, multiple skills share reference files, but no tests or verification results provided. Dependency availability: skills depend on external tools (e.g., Xcode, npx), but version requirements or failure handling not specified. Failure messages: no examples of error handling or user prompts. Deductions: lack of test evidence, dependency availability unclear, failure messages not addressed.
Evidence: README clearly identifies target audience (iOS/macOS teams) and lists multiple installation methods (skills.sh, Claude Code, Codex, etc.), adapting to different environments. Capability boundaries: README lists checks and skill purposes, but does not specify limitations (e.g., unsupported project types). Trigger precision: provides clear user prompt examples, but does not explain how skills are triggered. Environment fit: provides multiple installation options, but does not specify system requirements (e.g., Xcode version). Deductions: capability boundaries and trigger precision descriptions are not detailed.
Evidence: README is well-structured with quick start, how it works, checks, installation options, etc., good information architecture. Installation notes are detailed, offering multiple methods. Naming stability: skill names are consistent, but no version history provided. Examples and FAQ: provides community results table, but no FAQ. Known limitations: not explicitly listed. License: MIT license clearly stated. Versioning and changelog: release workflow exists, but no CHANGELOG provided. Maintenance responsibility: author is clear, and contribution guidelines exist. Deductions: missing known limitations and changelog.
Evidence: Output is an optimization plan file, shareable and reviewable, high output usability. Marginal value: provides over 40 checks targeting common build performance issues, clear value. Cost-benefit: low installation and usage cost, but requires manual approval and verification, which may be time-consuming. Deductions: cost-benefit assessment based on static analysis, not actual execution.
Evidence: README references Apple documentation and WWDC, but does not provide specific links. Community results table provides external links, but not verified. Fact-inference separation: README distinguishes checks from community results, but does not clearly indicate which are inferences. Deductions: references lack specific links, community results unverified, fact-inference separation not clear.
- Publisher identity is unverified; treat source with caution.
- Data flow transparency is insufficient; how scripts handle project data is not explained.
- Sensitive data handling is not mentioned; may access project files.
- Dependency security is not assessed; check dependency list.
- Rollback mechanism is incomplete; relies only on re-benchmarking.
- Community results are unverified; may overstate effectiveness.
What does this agent do, and when should you use it?
This is an open-source Agent Skill package for analyzing and optimizing Xcode build performance. It consists of six skills: an orchestrator (xcode-build-orchestrator) coordinating five specialist skills: benchmarking, compilation analysis, project analysis, SPM analysis, and build fixing. These skills work by benchmarking clean and incremental builds, auditing build settings, detecting compile hotspots, and analyzing Swift package dependencies. All analysis follows a recommend-first workflow, with no project files modified until the user explicitly approves. The optimization plan is generated at .build-benchmark/optimization-plan.md. The repository also includes a shared support layer with scripts, references, and schemas. Community results show incremental build improvements ranging from 5% to 87% across various apps.
The Agent Skill executes an end-to-end build optimization workflow. First, the xcode-build-orchestrator runs xcode-build-benchmark to measure clean and incremental build times, producing timestamped artifacts. Then, xcode-compilation-analyzer analyzes compile hotspots, xcode-project-analyzer audits build settings, schemes, script phases, and target dependencies, and spm-build-analysis examines the package graph. These analyses converge into an optimization plan listing over 40 checks, including build settings audit, script phase analysis, compile hotspot detection, zero-change build overhead, target dependency review, module variant detection, SPM graph analysis, Swift macro impact, SwiftUI view decomposition, asset catalog parallelism, access control optimization, and incremental build diagnostics. After the plan is generated, the user reviews and approves items, and the xcode-build-fixer applies changes and re-benchmarks to verify improvements. The flow never modifies project files until approval.
- iOS or macOS developers facing slow incremental builds in their local development loop want to identify and eliminate bottlenecks.
- Teams investigating a recent build-time regression need evidence-backed optimization instead of guesswork.
- Project owners with extensive script phases or complex target dependencies want an audit of build settings against best practices.
- Developers seeking long-term build performance monitoring across teams, machines, or Xcode versions can combine this with RocketSim Build Insights.
- Developers in headless or CI environments want to automate benchmarking and optimization workflows without manual configuration.
What are this agent's strengths and limitations?
- Provides an end-to-end recommend-first workflow that never modifies project files before approval.
- Covers over 40 checks across build settings, scripts, compilation, and Swift packages, giving a comprehensive analysis.
- Community results demonstrate real improvements, such as an 87% reduction in incremental build time for one app.
- Produces a shareable evidence file (optimization-plan.md) that aids collaboration and PR reviews.
- Requires npx and Node.js for the skill installer, which may be a barrier for non-Node users.
- The skills may not fully cover non-standard build configurations or custom scripts outside common patterns.
- The orchestrator requires all skills to be installed, increasing the installation footprint.
- Actual gains vary by project, and some improvements may require manual tweaking after the agent's suggestions.
How do you install or deploy this agent?
Install all six skills via npx: npx skills add https://github.com/avdlee/xcode-build-optimization-agent-skill. Alternatively, install a single skill using the --skill flag, e.g., npx skills add https://github.com/avdlee/xcode-build-optimization-agent-skill --skill xcode-project-analyzer. Other options include the Claude Code plugin marketplace, Codex/OpenAI-compatible tools (copy skill folders to $CODEX_HOME/skills/), and the pi package manager. For manual installation, clone the repository and symlink the skill folder to your AI tool.
How do you use this agent?
Open your Xcode project in your AI coding tool and issue the command: Use the /xcode-build-orchestrator skill to analyze build performance and come up with a plan for improvements. The agent will benchmark clean and incremental builds, audit settings, find compile hotspots, and generate an optimization plan at .build-benchmark/optimization-plan.md. Review the plan, check the approval boxes for items you want, and ask the agent to implement: Implement the approved items from the optimization plan at .build-benchmark/optimization-plan.md, then re-benchmark to verify the improvements. Ensure all six skills are installed, as the orchestrator depends on the others.
How does this agent compare with similar options?
This repository complements RocketSim Build Insights, which provides long-term performance monitoring. It also coexists with other agent skill packages like Swift Concurrency Expert and SwiftUI Expert, which focus on specific development domains rather than build performance.