Syke provides a local-first memex (memory index) for agents, exposing commands to read the current memex, run evidence-backed queries, record observations, and perform health checks. It helps agents maintain continuity across sessions and harnesses by keeping a shared, up-to-date memory store.
Use Syke when an agent should consult recent session context before acting, when you want to record important observations for future runs, or when debugging agent memory/state. Ideal for workflows that require continuity between interactive sessions or background agents.
syke memex, use syke ask for deeper queries, run syke record to persist observations, and run syke doctor for diagnostics. Also covers pitfalls and provider configuration.Terminal/CLI-capable agents and installers (agents that can run pipx or shell commands). Works best where the agent can access local files and run the syke binary.
Syke is a local-first cross-harness memex for AI agents providing query, read, and record capabilities via CLI. The SKILL.md is thorough with clear command reference, exit codes, pitfalls, and setup instructions for both interactive and agent modes. Scripts are developer/QA tooling (release gates, smoke tests, Docker QA) — not user-facing — and most require specific runtime dependencies (uv, Docker, systemd) to execute fully. The one script that ran successfully (fresh-install-test.sh) demonstrated safe dry-run behavior by default.
uvdockersystemd --userWell-crafted skill with professional-grade release engineering scripts. No security concerns — no credential leakage, no suspicious network calls, no unguarded destructive commands. The SKILL.md follows AgentSkill spec closely with frontmatter, clear procedure, pitfalls, and verification steps. Architecture is exemplary with good separation between user-facing commands and internal scripts.