semitexa / llm
Self-hosted LLM assistant for Semitexa Framework — console-based skill execution
Requires
- php: ^8.4
- semitexa/cache: *
- semitexa/core: *
- semitexa/prompt: *
Requires (Dev)
- phpunit/phpunit: ^10.0
Suggests
None
Provides
None
Conflicts
None
Replaces
None
- dev-master
- 2026.09.10.0641
- 2026.09.08.2003
- 2026.09.05.1548
- 2026.09.02.1207
- 2026.08.25.0718
- 2026.08.04.2028
- 2026.07.16.2213
- 2026.07.16.2136
- 2026.07.16.0803
- 2026.07.14.0610
- 2026.07.12.0729
- 2026.07.09.1540
- 2026.07.06.1438
- 2026.06.21.0352
- 2026.05.08.1640
- 2026.04.09.1145
- 2026.04.08.2025
- 2026.04.06.1418
- 2026.04.05.0855
- 2026.04.03.1349
- 2026.04.03.1240
- v1.0.4
- 1.0.3
- v1.0.2
- dev-develop
This package is auto-updated.
Last update: 2026-09-10 06:49:48 UTC
README
Self-hosted LLM assistant with console-based skill discovery and execution.
Purpose
Integrates LLM capabilities into the Semitexa CLI. Console commands marked with #[AsAiSkill] are discoverable by AI agents with structured metadata including risk assessment and execution policies.
Role in Semitexa
Depends on semitexa/core for attribute discovery and the console command system. Skills are registered via ClassDiscovery and exposed to LLM agents as a structured manifest with execution constraints.
Key Features
#[AsAiSkill]attribute for skill discoverySkillRegistrybuilds manifest from ClassDiscovery- Execution policies:
AiExecutionKind,AiRiskLevel,AiConfirmationMode - Structured skill metadata for LLM agents
- Ollama integration for local inference
Notes
Optional package. Skills are only exposed when an LLM provider is configured. Execution policies ensure destructive operations require explicit confirmation.