melodic-software
- 406 skills
- 0 followers
- 30 repo stars
- 5 days ago last updated
- ▌ Clean · melodic-software bundleAudit an arbitrary directory tree for orphaned, temporary, stale-lock, failed-write, partial-download, and empty leftover artifacts; classify evidence into confidence tiers; and optionally remove exact validated paths after explicit per-tier approval. Read-only by default and manual-only. Use when: 'audit this directory', 'find orphaned files', 'what junk can I clean up', 'reclaim disk space', 'find temp or lock leftovers', 'clean up my home directory'. Skip when: repository cache/build cleanup belongs to repo-hygiene, a product has its own prune/GC command, or the target is an OS-managed root.
- ▌ Target · melodic-software bundleIdentify and rank optimization targets by EVIDENCE QUALITY rather than by suspicion, so an unmeasured system yields 'instrument this first' instead of a guess. Accepts targets from the current session's own pain, a named path or component, a telemetry store, or an open-ended 'what is slow here'. Ranks each candidate by how well its cost is actually attributed, names the drift-immune counter that would settle it, and refuses to rank an unmeasured candidate above a measured one however plausible its mechanism. Use when: choosing what to optimize, locating a bottleneck, or judging whether a suspected target is worth the work: 'what should we optimize', 'find the bottleneck', 'this feels slow', 'is X worth optimizing'. Entry point for the measurement-first optimization workflow; hands off to /performance:goal. Skip when the target is already chosen and measured (go straight to /performance:goal), or when a specific failure needs root-causing rather than a candidate ranking (that is debugging).
- ▌ Verify · melodic-software bundleRe-derive a performance result in a FRESH CONTEXT that does not inherit the implementer's numbers, then report the target as met or not met without rounding a miss into a win. Dispatches a verifier told to distrust the reported figures and reproduce them from the trees, checks that the change did not alter behavior via a differential covering every MODE the subject runs in, and states any behavior change separately from and above the performance claim. Use when: independently re-deriving a measured performance claim before it is stated, or writing up the result: 'verify this speedup', 'is this result real', 'independent verification', 'write up the performance result'. Final phase; runs after /performance:snapshot post. Skip when no baseline exists (there is nothing to verify), or when reviewing a diff for general quality rather than checking a measured claim.
- ▌ Interview · melodic-software bundleInterview relentlessly to reach shared understanding on a plan, decision, or idea. Questions arrive in frontier rounds: every question whose prerequisites are settled asked together as one numbered set, each with a recommendation. Routes by context: an engineering task locks a task contract (goal, constraints, acceptance criteria, named assumptions) into a PLAN.md Brief that feeds the planning pipeline; a general decision drives to a shared understanding and stops. Synthesizes directly when intent is clear, runs the rounds loop when gaps remain, or interviews relentlessly on request. Use when: 'interview me', 'lock the brief', 'spec this task', 'grill me', 'this is underspecified', 'ask me questions first', 'what do you need to know', 'acceptance criteria', 'how will we know this is done', or proactively before behavior-changing work when intent is ambiguous or underspecified; skip for mechanical work (typo/lint/whitespace/rename) and casual conversation.
- ▌ Code Review · melodic-software bundleCI code-review lane for a GitHub pull request. High-signal correctness and maintainability findings only, scoped out of security when a security lane exists. Use when: 'CI code review', 'claude-review lane', '/review:code-review', or a reusable workflow invokes the org code-review plugin command.
- ▌ Retro · melodic-software bundleRun a structured session retrospective: extract transcript metrics, assess quality across five dimensions, check feedback-memory regressions, and codify learnings durably. Use when: 'retro', 'retrospective', 'what did we learn', 'how did I do', 'codify learnings', 'show trends', or at end of session; modes: session (default), codify, trends, quick.