Tracker Sync

GSDL's tracker write path. Takes an already-decided payload — slice titles, a comment body, a target state — and applies it to whichever tracker the project is configured for. This skill makes no judgment calls: it does not author comments, choose slice boundaries, or pick which items to close. Callers decide; this skill executes.

Tier

SMALL. Every operation here is: resolve an ID, POST a mutation, report the result. There is no reasoning to spend a larger model on, and tracker round-trips are frequent enough across the pipeline that the cost difference is real. The orchestrator spawns a SMALL subagent for each of these.

Non-interactive. Never ask the user anything. If credentials are missing, an ID won't resolve, or a mutation fails, return a BLOCKER describing exactly what failed and what the caller should do — including the exact text a human could paste into the tracker manually.

Provider Resolution