skills/avdlee__xcode-build-optimization-agent-skill

mirror of https://github.com/AvdLee/Xcode-Build-Optimization-Agent-Skill.git synced 2026-09-14 13:59:59 +08:00

Files

T

Antoine van der Lee cc642e0b52 Add Xcode build optimization skills, references, scripts, and plugin metadata

First versioned snapshot of the full skill repository including:
- Five installable agent skills under skills/
- Shared references, schemas, and helper scripts
- Plugin and marketplace metadata for Claude Code
- GitHub workflows for release and README sync
- Compilation caching as a recommended build setting
- Four deeper Swift compiler diagnostic flags for compilation analysis
- Fixed SWIFT_COMPILATION_MODE to use singlefile (actual build setting value)

2026-03-14 17:09:17 +01:00

2.7 KiB

Raw Blame History

GitHub Copilot Instructions: Xcode Build Optimization Skills

Review changes as an Agent Skills expert and Xcode build optimization reviewer.

Repository Mission

Keep this repository focused on:

Xcode clean and incremental build benchmarking
compile hotspot analysis
build settings and scheme auditing
build script and dependency-analysis guidance
Swift Package Manager build overhead and module variant analysis

Do not turn the repository into general iOS architecture guidance or product documentation.

Agent Skill Requirements

Every skill directory must contain a SKILL.md with valid name and description frontmatter.
Descriptions must state both what the skill does and when to use it.
Keep SKILL.md concise and delegate deeper material to one-level-deep reference files.
Reference files should be discoverable from the matching SKILL.md.

Behavior Constraints

Default to recommend-first behavior.
Never introduce instructions that apply project, build-setting, or source changes without explicit developer approval.
Preserve evidence-driven workflows: command used, build type, target or package affected, and measured impact.
Prefer deterministic shell commands and structured artifacts over ad hoc prose.

Build Optimization Focus Areas

When reviewing changes, prioritize:

Benchmark correctness
- clean vs incremental builds are treated separately
- warm-up and repetition rules are explicit
- output artifacts are stored in .build-benchmark/
Evidence quality
- recommendations cite timing summaries, script phases, compiler diagnostics, or dependency data
- estimated impact and confidence are separated from observed evidence
Apple-aligned guidance
- target dependencies and build order accuracy
- script input/output declarations and .xcfilelist usage
- DEFINES_MODULE, module maps, and framework-qualified imports
- explicit module dependency and module variant reduction guidance
- explicit type information and simpler expressions for Swift compile efficiency
Safety
- scripts should analyze and format data, not mutate user projects
- workflows should avoid surprising side effects

Common Issues to Flag

Vague skill descriptions with no trigger conditions
Overlong SKILL.md files that duplicate reference content
Advice that skips measuring before optimizing
Guidance that ignores incremental builds
Destructive automation or auto-fix behavior without approval
Inconsistent benchmark or recommendation fields across files

Useful References

references/build-optimization-sources.md
references/benchmark-artifacts.md
references/recommendation-format.md
schemas/build-benchmark.schema.json

2.7 KiB Raw Blame History