Skip to content

Roadmap

forkctl ships small, landed chunks. Nothing goes in the “soon” column that doesn’t already have a design doc.

v1.1.0 (2026-04-20) — Layer 7 (Rename). Three new tools taking the surface to seven layers and twenty-two tools. AST-aware polyglot rename across identity files, code symbols (JS/TS/TSX/HTML/CSS bundled; other languages resolve at runtime), non-code textual surfaces, and a post-pass that regenerates lockfiles and emits an asset-regeneration manifest. Snapshotted apply; one-command rollback. See the Rename handbook page for the walkthrough.

v1.0.0 (2026-04-19) — initial release. Six layers, 19 tools, all schema-validated and audit-logged. MCP server + CLI with full parity.

See the CHANGELOG for the full list.

Forking a repo is easy. Adopting it as your own product means every reference — package names, class identifiers, badge URLs, lockfile hashes, even binaries — needs to change coherently. Existing tools either did template-time substitution (fails on repos that weren’t built as templates) or sed -i chains (brittle, no casing awareness, corrupts partial matches).

Layer 7 shipped AST-aware rename in v1.1.0: a planner that understands what an identifier actually is and renames it the way a refactor tool would, while also handling every non-code surface (README, badges, repo URLs, lockfiles, binaries).

Tool surface (shipped in v1.1.0)

  • forkctl_rename_plan — returns a reviewable RenamePlan.
  • forkctl_rename_apply — consumes the plan, returns a RenameReceipt. Only mutating command.
  • forkctl_rename_rollback — restores from latest snapshot.

plan is read-only. apply requires a plan file produced by plan. There is no one-shot --yes flag — renames are consequential and the two-step flow is deliberate.

The full design — scope rules, per-language AST handling, snapshot/rollback contract, non-goals — lives in design/rename.md in the repo.

Nothing else has a design doc yet. If something here matters to you, open a discussion at https://github.com/mcp-tool-shop-org/forkctl/discussions — design work happens there first.

  • Fleet-wide rename. Once Layer 7 is solid, batch-apply a rename across many adopted forks.
  • Policy preflight for enterprise migrations. Today preflight-policy checks a single repo; enterprise adopters often want to assess a whole org.
  • Bootstrap profile: plugin-fork. A profile tuned for forks that are plugins/extensions of an upstream framework — keeps upstream sync but also keeps a clean contributor-facing surface.
  • File an issue with a concrete scenario, not just a wish. “Here’s what I’m trying to do / here’s where forkctl got in the way” beats “please add X.”
  • Design discussions happen in Discussions before any code or doc lands.
  • If you’re shipping a fork-heavy workflow at scale, open an issue with the shape of it — real adoption patterns drive what gets built next.