Read-only real-time explorer for AI coding-agent session snapshots (ai-agent v2/v3, claude-code, codex, opencode). Workstation-only, single static binary.
Skill Claude CodeCodex
Step-by-step workflow for adding, modifying, or debugging a source-format adapter. Use whenever editing an adapter under internal/adapters, internal/canonical, or related fixture files.
Skill Claude CodeCodex
Apply ai-viewer coding standards for production-quality Go and TypeScript changes. Use for implementation, refactoring, runtime behavior changes, and any change where clean code, separation of concerns, modularity, or maintainability matters.
Skill Claude CodeCodex
Helper-subagent patterns for ai-viewer. Use when bounded read-only investigation, summarization, or parallel context gathering would help. Do not use this skill to delegate implementation by default, and do not use helper subagents as external reviewer gates.
Skill Claude CodeCodex
Install, run, and operate ai-viewer on the operator's workstation — the system install (/opt/ai-viewer, runs-as-operator, explicit --source flags, systemd units), the user install alternative, build/run/upgrade/uninstall commands, and the threat-model reasoning behind the run-as-operator decision. Use when installing…
Skill Claude CodeCodex
Apply React/TypeScript/Vite patterns used in ai-viewer's frontend. Use when editing frontend/ — pages, components, API client, SSE handling, D3 visualizations.
Skill Claude CodeCodex
Apply Go-specific patterns used in ai-viewer for fsnotify watching, SQLite access, SSE streaming, and structured logging. Use when editing internal/ingest, internal/store, internal/presenter, internal/adapters, internal/canonical, or internal/obs.
Skill Claude CodeCodex
Catalog of every automated quality gate ai-viewer enforces — commands, thresholds, and what to do when a gate fails. Use before claiming any work done, before any commit, when adding a new gate, or when investigating a CI failure. The runtime companion to .agents/sow/specs/quality-gates.md.
Skill Claude CodeCodex
Run ai-viewer's external reviewer gates. Use on meaningful chunks of work, at least per SOW and per substantial SOW milestone: gap analysis, implementation plan, and implementation review. Reviewers are glm, minimax, kimi, mimo, deepseek, and qwen. They are gates, not implementers.
Skill Claude CodeCodex
Keep ai-viewer specs synchronized with code. Use whenever a code change touches runtime behavior, defaults, schemas, contracts, file layouts, or external interfaces.
Skill Claude CodeCodex
Run, write, and maintain ai-viewer tests across Go backend and React frontend. Use when adding or changing tests, debugging CI failures, managing fixtures, or running benchmarks.
Skill Claude CodeCodex
Master orchestration cycle for ai-viewer work — gap→plan→spec→test→code→gates→review→commit. The CTO writes implementation directly. Helper subagents are optional for bounded investigation, not the implementation norm. External reviewers are three milestone gates per meaningful feature, substantial batch, or SOW, not…