Evaluates Bazel adoption readiness and plans staged repository migrations with measured signals.
npx skills add https://github.com/ocx-sh/grimoire-lore --skill bazel-adopt-ocx-sh
Please help me install this Agent Skill. Skill: bazel-adopt Source: https://github.com/ocx-sh/grimoire-lore/tree/main/skills/bazel-adopt Command: npx skills add https://github.com/ocx-sh/grimoire-lore --skill bazel-adopt-ocx-sh
gh, generator maturity via the Bazel Central Registry, submodule coupling, team ownership), recorded in a decision file with no invented thresholds. - Staged migration procedure: Eleven ordered steps covering pilot selection, .bazelversion pinning, the 8.x-versus-9.x call, MODULE.bazel and lockfile setup with a day-one freshness gate, rc-file layering, four-stage cache wiring, and CI lane taxonomy. - Per-language branches: Dedicated migration branches for Rust (crate_universe, crates_vendor, prost), Python (rules_python, pip.parse, pytest wrappers), TypeScript (rules_js/rules_ts, pnpm-only ingestion), and C++ (toolchain constraint tree), each with its first three rules to satisfy. - Use Case: A team with a polyglot monorepo and 45-minute CI runs asks whether Bazel is worth it. The Skill measures their actual CI wall-clock, checks generator maturity per language, produces a written go decision, then migrates a Rust pilot with a pinned toolchain and a committed lockfile before touching anything else. ## Quick Start Ask the assistant to evaluate whether your repository should adopt Bazel and produce a migration plan using the bazel-adopt skill.High-intent search queries and answers about installing and using this skill.
Measure five signals instead of applying a threshold: median whole-repo CI wall-clock, whether the largest CI job is actually a build, generator maturity per language, cross-repository coupling, and build ownership. No published numeric cutoff exists, so the verdict names which measured signal decided it.
Follow an ordered procedure: pick a small hermetic pilot in one language, pin an exact Bazel version in .bazelversion, create MODULE.bazel with a committed lockfile and a day-one freshness gate, layer rc files, then wire the disk cache before any remote cache. Run both old and new builds as required checks throughout.
Start on the Active LTS, which is 9.2.0 as of the skill's verification date, unless a consumer pins the Maintenance major. Note that 9.x deletes all WORKSPACE logic, empties autoload so every py_*, sh_*, and cc_* symbol needs an explicit load(), and removes bazel sync and bazel analyze-profile.
Try the narrower fix first: sccache and cargo nextest for Rust, uv for Python, pnpm with Nx or Turborepo for TypeScript. The evidence steers Python-heavy monorepos toward Pants before Bazel and holds Nx or Turborepo sufficient for JS-only teams; Bazel fits multi-language repos where build time is the measured cost.
Exit 37 covers two distinct cases: an existing dependency's locked version changed (an internal IllegalStateException) or MODULE.bazel gained a new bazel_dep (the documented missing-checksum message). Gate CI on the exit code, never on matched stderr text, and never hardcode the lock schema version.
Remote execution comes last, only after a five-step readiness gate passes, and target selection is deferred until whole-repo median wall-clock consistently exceeds roughly 40 minutes. Both production selection tools document real false-negative classes, so whole-repo bazel test //... remains the default backstop.
danmartinez78Configures GitHub authentication for the Hermes agent via gh or git.
HermetiqAnalyze Bazel build telemetry to diagnose slow builds and cache misses.
buurgueesBlock merges or deployments failing code quality, migration, security, and documentation checks.
PriyanshKuniyalManage and optimize monorepo setups with Turborepo, Nx, and pnpm.