sanlee-ys
UserOperating layer for a multi-vendor agent fleet: security posture, mechanical guards, postmortems, and working agreements.
Categories
Indexed Skills (5)
dcb
Scaffold a task into the user's DCB framework (Direction, Contracts, Bar) before starting ambiguous, consequential, or hard-to-reverse work with Claude, and for consequential work also produce a short plan document that is reviewed before any implementation starts. Invoke via /dcb <task description>, or /dcb alone to be asked what task is being scoped. Use this when the user explicitly types /dcb, or says things like "let's DCB this", "set direction/contracts/bar for X", or asks to scope out a risky/ambiguous piece of work before diving in. Do NOT use for quick mechanical edits, renames, or bounded tasks the user has already fully specified — DCB is overhead there, not help.
handoff
Produce a paste-ready handoff brief capturing the LIVE state of the current work session, so another Claude Code window (which cannot read this one's transcript) or a future session can resume without losing the thread. Invoke via /handoff. Use when the user is pausing, switching windows/machines, ending a pairing session, or says things like "hand this off", "write a handoff", "I'm moving to the editor", "pick this up later". Distinct from DCB: DCB is a PRE-flight scaffold (Direction/Contracts/Bar set before work starts); handoff is a POST-flight snapshot (state captured to resume/transfer work already in progress). Do NOT use as a substitute for actually committing/pushing work, and skip it for trivial one-off questions where there is no in-progress state worth transferring.
proglog
Append a dated entry to the user's private pairing-journal file — concepts relearned (with an analogy to a prior stack when one genuinely fits) and what got built this session. Invoke via /proglog, or proactively offer it at the end of a substantive hands-on pairing/relearning session. Also proactively check at the START of a session that looks like a continuation of pairing work: if the latest prog-log entry looks like it predates work that's evidently happened since, remind the user before diving back in. Do NOT use for quick one-off questions, mechanical edits, or sessions with no relearning/building content worth journaling — and don't nag if the user declines the reminder once.
descope-sweep
Sweep every one of the user's repos for stale references to something that was cut, renamed, ported, or scrapped (a dropped dependency, a superseded metric, a language/framework port, a rejected experiment, a renamed contract) — covering not just the obvious docs but the long tail (ADRs, .env.example, KB stubs an agent might index and speak, the sources of generated docs, stale metric cells, SYS-NNN/ADR-NNN ranges). Invoke via /descope-sweep <old-term> <what-it-changed-to>, or when the user mentions something was cut/replaced/ported and wants references cleaned up. Do NOT use for a single-file rename inside one PR — just fix that inline. This is for changes that cross, or could plausibly cross, repo boundaries.
park
Append a stray project/portfolio idea to the user's private ideas parking-lot file, so it's captured without derailing current work or building it prematurely. Invoke via /park <idea>, or when the user says "park this" or floats an idea he doesn't want to build right now. Do NOT use for ideas that already have an obvious home (an existing repo's own backlog/issue tracker) or for fully-scoped work ready to start now — park is only for not-yet-ready stray ideas.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.