Keep your AI tools in sync with your project.
A single-file Python CLI that injects project context into every AI session — no lost state, no repeated setup, no invisible costs.
Every AI session starts from scratch. Switch from Claude to Gemini and it has no idea what you were building. Switch accounts and your task list evaporates. Costs accumulate invisibly.
init
Bootstrap project-docs/ and AI instruction files
exec
Run any AI CLI with context injected automatically
checkpoint
Archive completed tasks, refresh state, log telemetry
status
Dashboard: tasks, budget, alerts, watcher state
Context injection
Shared project-docs/ directory every AI tool reads at session start
Cost tracking
Budget limits with 80% warnings. Spend read from costs.md
Watch daemon
Background process regenerates context.md as project-docs/ changes
Sentinel alerts
Detects flatline failure loops — 3 consecutive failures of the same command
v0.3.0 — shipped
Enriched agent templates, AGENTS.md (Codex), init flags for org/repo/project-id.
v0.4.0 — next
Trio Bootstrap: synlynk run drives an Architect → Build → Verify pipeline across your configured agents.
v1.0 — the OS
Agent routing, job control, async dispatch, open context protocol, and a Tokq cloud bridge.