Discover Documentation skills, compare source repositories and read the instructions for each workflow.
[PHASE 4] Implements and validates fixes for verified theories. Entry: PERFORMANCE_THEORIES.md has a verified theory. Outputs: Code fix, BENCHMARK_BASELINE.md new column, theory marked fixed/fix-failed. Next: L...
[ORCHESTRATOR] Runs full performance optimization loop for Truffle languages. Entry: User wants to optimize performance. Process: Determines current state, executes appropriate phase (1→2→3→4), loops until perf...
Experimental allocation profiler tracking memory allocations at guest-language level. Shows allocation sites, object types, and memory pressure patterns. Use to identify allocation-heavy locations, find unneces...
Detects optimization barriers during Truffle compilation. Identifies virtual calls, non-constant stores, unresolved type checks, and TruffleBoundary issues that prevent peak performance. Reports exact source lo...
Dumps and analyzes Graal IR compiler graphs showing optimization decisions. Use BGV format with bgv2json/seafoam to inspect escape analysis, boxing removal, inlining decisions, and call node types. Reveals what...
[PHASE 2] Analyzes code to identify optimization opportunities. Entry: BENCHMARK_BASELINE.md exists AND (no PERFORMANCE_THEORIES.md OR all theories fixed). Outputs: PERFORMANCE_THEORIES.md. Next: PHASE 3 (verif...
[PHASE 3] Proves theories with profiling tools. Entry: PERFORMANCE_THEORIES.md has unverified theories. Outputs: Theory marked verified/falsified. Next: PHASE 4 (implementing-performance-fixes) when one theory...
Logs every compilation event with timing, tier (T1/T2), success/failure, and invalidation reasons. Use to verify hot code is compiling, diagnose compilation failures/bailouts, track recompilation cycles, and un...
[PHASE 1] Creates benchmark suite and performance expectations. Entry: No BENCHMARK_BASELINE.md exists. Outputs: BENCHMARK_BASELINE.md, benchmark files. Next: PHASE 2 (generating-performance-theories).
Generate English commit message with emoji prefix and push. Use when user wants to commit and push changes, or uses /commit.
Comprehensive YAML syntax validation, error fixing, and schema validation for various formats (GitHub Actions, Docker Compose, Kubernetes, GitLab CI). Use when Claude needs to: (1) Validate YAML syntax, (2) Che...
Creates professional PowerPoint presentations from content using HTML/CSS slides. Can also read and analyze existing PPTX files. Use when the user asks to create slides, build a presentation, generate a deck, m...
Analyze conversation to extract reusable engineering rules. Use when user says 'reflect', 'remember this', 'learn from this', or after receiving corrections/feedback.
Guidelines for identifying the appropriate AI persona (Flash, Sonnet, Pro) for a given task.
Gather design feedback from AI consultants (Gemini and Codex). Use for architecture decisions, design review, or when you want multiple expert perspectives on SENTINEL development. (project)
Deploy Codex or Gemini as working agents to implement tasks. Use when you want another AI to actually write code, not just review. (project)
Run security scans and vulnerability analysis using Gemini's security extension. Scan dependencies, generate PoCs, and document findings.
Validate wiki integrity using Kimi. Checks broken wikilinks, missing frontmatter, orphan pages, and structural issues in the Obsidian vault.
Generate character portraits using Gemini NanoBanana. Reads from character YAML and calls gemini directly with full guardrails.
Audit SENTINEL game data integrity. Validates regions, jobs, vehicles, and favors against schema enums.
Generate full NPC profiles from character YAML and save JSON to sentinel-agent/generated_npcs. Native (no external agent call).
Generate full NPC profiles using Codex. Reads from character YAML and generates agenda, disposition modifiers, dialogue, secrets, and hooks.
Use when you have a plan file from Claude Code plan mode and want to execute it autonomously - transforms plan markdown into execution config JSON for the mizu CLI with proper task extraction, permission infere...
Use when checking the status of an autonomous mizu plan execution - reads state files to show progress percentage, current task, completed tasks, and recent execution notes
The default order uses GitHub repository stars from our indexed data. Stars belong to the source repository and may be shared by multiple skills. They indicate repository interest, not skill installs or an independent quality score.