fivol

fivol/claude-make-better

Plugin Claude Code

Plugin marketplace listing 2 plugins: make-better, feature.

fivol/claude-make-better

Plugin Claude Code

Give every task its own isolated working copy — git worktree per repo, optional unique port, detached dev servers, pretty .localhost URLs and a live admin dashboard — then ship it through a review gate into a PR and merge it. Four composable skills (workspace, ship, pr-feedback, merge) plus an impartial review gate, a.

fivol/claude-make-better

Agent

Read-only investigator for the feature:review gate, deep tier. Runs exactly one reasoning-shaped job — a correctness angle, the altitude angle, a correctness/P0 verification batch, or the cross-repo pass — over a diff it did not write, and returns findings as data. It has no Edit/Write and never changes code. The…

fivol/claude-make-better

Agent

Read-only investigator for the feature:review gate, light tier. Runs exactly one retrieval-shaped job — one review angle, one cleanup verification batch, or the sweep — over a diff it did not write, and returns findings as data. It has no Edit/Write and never changes code. The review skill dispatches it; don't pick it…

fivol/claude-make-better

Command

Open the admin dashboard — one live view of all task workspaces.

fivol/claude-make-better

Command

Preflight-check the workspace toolchain (git, gh, config, repos, Caddy) and fix/instruct.

fivol/claude-make-better

Skill Claude CodeCodex

Land a finished task branch in its base branch — sync the base into the branch and resolve conflicts there, run the final review on the integrated code, wait for green CI, merge, then tear the workspace down (servers, worktree, branches, ports, proxy). Use ONLY on an explicit user go-ahead ("done", "merge it"…

fivol/claude-make-better

Skill Claude CodeCodex

Work the reviewer's comments on a pull request — list what is still unaddressed, act on each item (fix it, answer it, or argue with it), and post the replies back so nothing is silently dropped. Use when the user says to address / answer / go through the PR comments or review feedback, when you are about to add to a…

fivol/claude-make-better

Skill Claude CodeCodex

Deep adversarial review of the change in flight — correctness bugs first, then reuse/simplification/efficiency/altitude/conventions, plus git history, prior review comments and cross-repo integration — every candidate independently verified, then fixed in place. Use before a change is committed, and again on the whole…

fivol/claude-make-better

Skill Claude CodeCodex

Ship the change that is in the working tree — simplify it, run the review gate that finds and fixes its bugs, commit, push, open or update the PR, and report what landed. Use when the user asks to ship / land / commit / push the current work or open a PR for it ("ship it", "commit this", "оформи PR"), and when the…

fivol/claude-make-better

Skill Claude CodeCodex

Give a task its own isolated working copy of the product — a git worktree per repo off the configured base branch, dependencies symlinked, and (in full mode) a unique port, a detached dev server and a pretty http:// .localhost URL — then drive the task through build → ship → merge. Use proactively at the start of a…

fivol/claude-make-better

Plugin Claude Code

Continuously audit and improve an existing codebase. Ships /make-better (one-command flow) plus /systems-discover and /systems-review for fine-grained control. Audits across 9 topics: bugs, completeness, DRY, architecture, consistency, efficiency, tests, docs-sync, and security. Also ships /meta-review — a…

fivol/claude-make-better

Skill Claude CodeCodex

One-command Make Better — auto-decides between bootstrap, refresh, or review-only based on registry state. Orchestrates /systems-discover and /systems-review. Flags: [count] [subsystem] [--no-discover|--rebuild|--discover-only] [--yes].

fivol/claude-make-better

Skill Claude CodeCodex

Retrospective audit of the user's own Claude Code sessions. Use whenever the user wants to review, analyze, or reflect on their Claude Code chat history / logs / past sessions to surface recurring patterns, repeated mistakes, instruction violations, workflow friction, missing skills or automation, or suboptimal tech…

fivol/claude-make-better

Skill Claude CodeCodex

Companion to /systems-review. Builds and maintains docs/SYSTEMS.md by scanning the codebase, proposing systems grouped into sections, and merging cross-cutting systems across subsystems. Three modes: incremental (default), scoped by area, and --rebuild. Plan mode shows a compact list of system names for…

fivol/claude-make-better

Skill Claude CodeCodex

Periodic audit of codebase subsystems. Picks N stale systems from docs/SYSTEMS.md, runs topic-driven review (bugs, completeness, dry, architecture, consistency, efficiency, tests, docs-sync, security?), shows a per-system plan in plan mode, and applies fixes in isolated worktrees before stamping lastreview. Invoke as…