Thomashighbaugh/llm-cache
skills/llm-cache/SKILL.md
Semantic LLM response cache — caches and retrieves LLM responses for semantically similar prompts using vector similarity
$ install --global
skillsauthnpx skillsauth add Thomashighbaugh/opencode llm-cacheInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Jun 16, 2026, 2:28 AM178.1s1 file scanned
SKILL.md
- name:
- llm-cache
- description:
- Semantic LLM response cache — caches and retrieves LLM responses for semantically similar prompts using vector similarity
- level:
- 2
LLM Cache — Semantic Prompt Caching
Uses the existing vector index infrastructure (sqlite-vec) to cache LLM responses for semantically similar prompts. When a prompt is semantically similar (cosine similarity > 0.92) to a previously cached prompt, the cached response is returned instead of making a fresh LLM call.
How It Works
- On LLM call: Compute embedding of the prompt (minus dynamic context like file contents)
- Query vector index: Search for similar cached prompts
- Cache hit (similarity > 0.92): Return cached response
- Cache miss: Execute LLM call, cache the response
Storage
- Vector index:
.opencode/.vector/(sqlite-vec, gitignored) - Cache entries:
.opencode/cache/llm/{hash}.json - Embedding model:
Xenova/all-MiniLM-L6-v2(shared with vectorize-context)
Cache Key Structure
Cache keys are derived from the structural prompt — the prompt with dynamic context (file contents, error messages, timestamps) stripped. This means:
- "List subcommands for /ideation" → same key every time
- "Explain how React hooks work" → same key every time
- "Fix this error: [specific error]" → different key per error
TTL
- General prompts: 1 hour
- Static/system prompts: process lifetime
- Structural queries (subcommand listings, help text): 24 hours
Usage
This skill is invoked automatically by the cache system in plugins/hubs-plugin.ts. It can also be triggered manually:
/harvest-context cache llm "your prompt here"
Related
cache-utils.ts— Shared cache infrastructurevectorize-context— Vector index management/harvest-context cache— Cache management commands
Related Skills
Thomashighbaugh/tool-creator
tools
VerifiedTrustedCommunity
Create valid, type-safe TypeScript tools for OpenCode — generates correct boilerplate, typed interfaces, and handler stubs. Use when building new tools for global config or per-project .opencode/tools/.
1SKILL.mdUpdated Jul 5, 2026
Thomashighbaugh/tree-of-thoughts
testing
VerifiedTrustedCommunity
Explore multiple solution branches in parallel, evaluate each, and recommend the best path. Use when the user explicitly invokes /ideation tree-of-thoughts or asks for "tree of thought" / "branching exploration".
1SKILL.mdUpdated Jul 1, 2026
Thomashighbaugh/self-consistency
testing
VerifiedTrustedCommunity
Run multiple independent reasoning passes and find consensus. Use when the decision is critically important, the stakes are high, or the user explicitly requests "run it multiple times" / "check consistency" / "self-consistency".
1SKILL.mdUpdated Jul 1, 2026
Thomashighbaugh/opro
testing
VerifiedTrustedCommunity
Optimization by PROmpting — generate candidate prompt variations, test each against a benchmark, and report the best performer. Use when the user invokes /ideation opro or asks for "prompt optimization" / "OPRO".
1SKILL.mdUpdated Jul 1, 2026