cqlite

Resolve PR review comments for an in-review issue — fix them in the issue's worktree, re-run the gate + C as needed, push, and reply per thread. Fourth stage of the CQLite delivery pipeline. Use when the owner leaves PR comments and says "address

Turn a rough idea into one scoped, labeled GitHub issue (one P0–P3 + status:ready, testable acceptance criteria), and decide oracle-vs-design routing. First stage of the CQLite delivery pipeline. Use when the owner says "groom <idea>", "file an issue for X", or brings a rough idea to scope.

Implement an approved issue — spawn sstable-developer (TDD) in the issue's worktree, run rust-reviewer + roborev on the lite-green diff BEFORE any full gate (review-first is default), open the PR, then hand the endgame (the ONE full gate → C → final roborev → merge-on-green → finalize) to a disposable flow-closer agent so gate/review churn never accretes in the lead session. Third stage of the CQLite delivery pipeline. Requires owner approval of the spec (design-driven) first. Use when the owner says "implement

Pre-push validation checklist (cargo fmt, clippy with zero warnings, feature flag testing, test suite), CI monitoring, merge process, and release quality gates. Use when preparing to push code, validating changes before PR, running CI checks, merging PRs, or preparing releases.

Report status across all in-flight CQLite delivery work — render the shared GitHub Project claim board (item, status, assignee, priority) with drift + abandoned-claim reconciliation, then surface and drive the single furthest-along item waiting on the owner (a green PR to merge, a spec to approve), or offer a claim-aware pick-list. Use when the owner asks "where do things stand", "what's next", "unblock me", or "what needs me".

Activate a groomed design-driven issue — create its worktree + branch (1:1:1:1), run OpenSpec propose to produce proposal/design/specs/tasks, render the spec + recommended design INLINE, then STOP for the owner's approval. This is Seam 1; it never implements. Second stage of the CQLite delivery pipeline. Use when the owner says "activate

Finalize a merged issue — archive its OpenSpec change (sync delta specs into openspec/specs/), remove the worktree + branch, and close the issue with a traceable comment. Fifth stage of the CQLite delivery pipeline. Runs after the PR is merged — workers merge autonomously on green (gate PASS + C PASS + roborev clean), so finalize normally follows the worker's own merge. Use right after a merge (or when the owner says "finalize

Pre-push validation checklist (cargo fmt, clippy with zero warnings, feature flag testing, test suite), CI monitoring, merge process, and release quality gates. Use when preparing to push code, validating changes before PR, running CI checks, merging PRs, or preparing releases.

Implement and deserialize all CQL types including primitives (int, text, timestamp, uuid, varint, decimal), collections (list, set, map), tuples, UDTs (user-defined types), and frozen types. Use when working with CQL type deserialization, schema validation, collection parsing, UDT handling, or type-correct data generation.

Node.js bindings for Rust libraries using napi-rs. Use when working on Node.js/JavaScript bindings to Rust code, including creating or modifying napi macros and structs, converting types between Rust and JavaScript, handling errors across the FFI boundary, async/Promise patterns, building native modules, publishing to npm, testing binding code, or debugging binding issues. Tuned for CQLite (Cassandra CQL bindings) with feature parity tracking.

Python bindings for Rust libraries using PyO3 and maturin. Use when working on Python bindings to Rust code, including creating or modifying pyfunction/pyclass definitions, converting types between Rust and Python, handling errors across the FFI boundary, managing the GIL and memory, building wheels with maturin, publishing to PyPI, testing binding code, or debugging binding issues. Tuned for CQLite (Cassandra CQL bindings) with feature parity tracking.

Zero-copy deserialization, async I/O patterns, lifetime management, memory-efficient parsing, and safe handling of unsafe code for SSTable parsing. Use when working with performance optimization, memory efficiency, async/await, borrowing/lifetimes, zero-copy patterns, or memory usage under 128MB target.

Guide parsing of Cassandra 5.0+ SSTable components (Data.db, Index.db, Statistics.db, Summary.db, TOC) with compression support (LZ4, Snappy, Deflate). Use when working with SSTable files, binary format parsing, hex dumps, compression issues, offset calculations, BTI index, partition layout, or debugging parsing errors.

Generate real Cassandra 5.0 test data using Docker containers, export SSTables with proper directory structure, validate parsing against sstabledump, and manage test datasets. Use when working with test data generation, dataset creation, SSTable export, validation, fixture management, or sstabledump comparison.

DEPRECATED — superseded by flow-board. Kept as a pointer so old references don't 404. Use flow-board for the claim board + the one next thing.

DEPRECATED — superseded by the flow-* delivery pipeline. Kept as a pointer so old references don't 404. Use flow-groom / flow-implement / flow-board instead.

Rank open GitHub epics and issues by priority and recommend what to work on next. Read-only — never modifies issues. Invoke as /prioritize (optionally scoped to a label, milestone, or epic).

Comprehensive Rust coding guidelines with 265 rules across 26 categories. Use when writing, reviewing, or refactoring Rust code. Covers ownership, error handling, async patterns, concurrency, unsafe code, API design, memory optimization, performance, numeric safety, conversions, serde, pattern matching, macros, closures, observability, testing, and common anti-patterns. Invoke with /rust-skills.