Clarit-AI

Develop and contribute to the Plexium codebase — package structure, build commands, test patterns, architecture layers

Operate within a Plexium-powered repository — read wiki before work, update wiki after changes, follow ownership rules, use retrieval

Use when the user wants to install or verify the Plexium CLI from inside Codex.

Use when the user wants to initialize Plexium, set up the current repository for Codex, verify readiness, or connect Plexium's MCP server in Codex.

Use when the user wants to install Plexium, set up the current repo for Claude Code, verify readiness, or query the Plexium wiki from Claude Code.

Use when the user wants to search or retrieve information from the Plexium wiki in the current repository.

This skill should be used when the user asks to "configure Codex CLI", "edit config.toml", "set up Codex settings", "change the Codex model", "configure approval policy", "set sandbox mode", "add an MCP server to Codex", "create a Codex profile", "configure model providers", "set up OTel for Codex", or mentions Codex CLI configuration, config.toml, or settings. Provides comprehensive configuration reference for OpenAI Codex CLI.

This skill should be used when the user asks to "create a Codex hook", "configure Codex CLI hooks", "add a PreToolUse hook for Codex", "block commands in Codex", "create a stop hook", "set up SessionStart hooks", "write Codex hooks.json", or mentions Codex CLI hook events (SessionStart, PreToolUse, PostToolUse, UserPromptSubmit, Stop). Covers the Codex CLI hooks extensibility framework, configuration, event types, input/output wire format, and common patterns.

Triggers when the executing agent receives a markdown brief matching the 5-part format (Objective / Context / Steps / Expected outputs / Report back), or when the user says "follow the brief," "use the runbook contract," "report back per contract," or when a cross-machine handoff is in progress ("push branch," "fetch first," "consumer brief"). Applies to all work in the engram repo. Does NOT trigger on conversational requests that do not include a brief block.

Engram (formerly sglang-mamba) — persistent stateful inference for Mamba/SSM hybrid models on SGLang. Snapshot save/restore, 3-tier memory hierarchy, conversation_id tracking, tested across Granite, Nemotron, and Qwen architectures on H200. Use when working on the Engram repo (formerly sglang-mamba), debugging snapshot pipeline, deploying Mamba models, or running test phases.

Engram (formerly sglang-mamba) — persistent stateful inference for Mamba/SSM hybrid models on SGLang. Snapshot save/restore, 3-tier memory hierarchy, conversation_id tracking, tested across Granite, Nemotron, and Qwen architectures on H200. Use when working on the Engram repo (formerly sglang-mamba), debugging snapshot pipeline, deploying Mamba models, or running test phases.

Reference for the Vercel AI SDK Language Model Specification V4. This skill should be used when implementing or modifying a custom provider for the Vercel AI SDK, understanding the LanguageModelV4 interface, mapping prompts to provider format, implementing streaming, or handling errors in a provider context. Also covers the upgrade path from V1 to V4.