Claude Skills by bravros

A test fixture skill for verifying Claude model reference stripping in converted output. Triggers on "/fixture-model-req" or "run fixture".

Fetch PR review comments, implement the fixes, and push.

Research and advise how a software project should be built — stack selection, architecture, operating-cost tradeoffs, and real-world comparables. Use on `/advise-project-approach` or when asked for project strategy, stack choice, or an architecture/repo critique.

Generate an after-merge.md operational deploy checklist for homolog-to-main aggregate releases. Use when the user mentions deploying multiple PRs to production, needs a post-merge runbook, or asks what to do after merging to main.

Fully autonomous SDLC pipeline — plan to PR, zero user intervention. Invoke via /auto-pr.

Capture, list, and promote pre-planning ideas. Use `/backlog` to add, view, promote, complete, or drop ideas before planning.

Verify N PRs, address review feedback, ordered-merge to the staging branch (never main), close linked issues/backlog per merge, then hand off the full test suite. Use on /batch-merge-prs.

Commit staged changes with emoji+type conventions and formatting.

Scan project and generate/audit CLAUDE.md files with stack auto-detection and Context7 docs. Use when the user says /context, generate context, audit CLAUDE.md, update project instructions, or needs AI context files refreshed.

Runs Claude Code's built-in /doctor health check, then audits the workspace against Anthropic's 6 then-and-now context-engineering shifts (rules→judgement, examples→interfaces, progressive disclosure, one-home instructions, auto-memory, rich references). Reports first, fixes only on approval. Use on /doctor-plus or "context checkup".

Complete a feature — merge the approved PR, record plan completion, route the homolog→main decision. Use on `/finish` or "finish the feature".

Bootstrap a private GitHub repo for the current folder and wire the origin remote. Invoke via /git-this.

Report knowledge-graph label coverage across every graphify-enabled project on this machine — communities, labels, and how many nodes still render as "Community NN". Use on `/graphify-status`, or when asked which graphs are stale, unlabelled, or degraded.

Set up a graphify knowledge graph for the current project — in-project committed graph.json, AST extraction + external labelling, tracked refresh hooks, union-merge driver, user-scoped MCP query surface. Use on `/graphify-this-project`.

Emergency hotfix deploy — commit, push homolog, PR to main, merge now. Use on `/hotfix <description>`.

Stress-test a plan or design via a round-by-round interview — ask the whole frontier of independent questions each round until every branch of the decision tree is locked. Use on `/interview-me`, "interview me", or when the user signals doubt about a plan.

Prepare a Laravel app for Laravel Cloud — SQS FIFO message groups, edge-cacheable public pages, Horizon removal. Use on `/laravel-cloud-ready`, or when a deploy hits MissingParameter MessageGroupId.

Adopt Pest's Rector plugin on a Pest 5 project and refactor tests into Pest matchers — scoped to tests/, dry-run gated. Use on /laravel-pest-rector, or when adding pest-plugin-rector.

Upgrade a PHP/Laravel repo from Pest 4 to Pest 5 and turn on Test Impact Analysis (TIA) so local full runs only re-execute the tests a change can reach. Use this whenever someone mentions upgrading Pest, moving to Pest 5, `pestphp/pest ^5`, Test Impact Analysis, `--tia`, "why is my test suite so slow", speeding up a slow PHP test suite, `Tia mode requires Pest tests`, `WorkerCrashedException` after enabling TIA, converting PHPUnit class-style tests to Pest, or setting up a TIA baseline — even...

Run a local PR review without the @claude GitHub Action.

Store, read, inject, and rotate secrets via the 1Password CLI (op). Enforces op:// references so secrets never touch env files. Use when the user mentions 1Password, op:// secrets, credential rotation, or safe secret injection into code.

Plan and run implementation from a .planning findings dossier — you derive units, waves and model tiers for maximum parallelism, subagents write the code, you verify diffs and commit. Use on /orchestrate [folder].

Post @claude review comment on the current PR and ask what's next. Use on `/pr-review` to trigger the GitHub Actions review workflow.

Create a Pull Request with plan context and base branch detection.

Eliminates generic AI slop from React/Next.js frontends with premium typography, color calibration, and motion choreography. Triggers on /premium-website or requests for anti-slop design.

Fast `homolog → main` merge for committed, pushed work. Trigger — `/promote`. Requires out-of-band token minted via `bravros promote unlock` from a separate terminal — Claude cannot mint it.

Safely prune already-merged branches (local + remote) with 7-day tombstone recovery. Manual-only — nothing auto-triggers it. Invoke via `/prune-merged`.

Push current branch to remote with branch safety checks.

Quick task execution without a full plan — just do it and commit.

Investigate a bug report or feature request and document it as ONE multi-file .planning dossier for /orchestrate to plan from. Use on `/recon <problem or feature>`, `/recon --worktree`, or `/recon B-NNNN`.

Investigate a defect with graphify and code references, then certify the root cause with runtime proof. Never modifies code. Runs standalone on `/scout <bug>`, or as the defect arm of `/recon`.

Commit and push changes in one step with safety checks.

EXPLICIT-INVOCATION ONLY — trigger only when the user types /start. Initializes a new project with stack-aware CLAUDE.md, .bravros.yml, .gitignore, and base structure. Do NOT trigger on natural-language phrases like init or setup without the slash.

Read-only drain of a stale issue + backlog queue — dedup, classify each item vs LIVE code (already-done / partial / superseded / no-longer-needed / open / human-only), adversarially verify every close, then apply closes/cancels serially. Use on /triage-sweep.

Update git hooks in an existing project to the latest version from ~/.agent_config/templates.

Health-check the Bravros SDLC install — skill drift, config, hooks, toolchain — and optionally repair it. Use on `/verify-install`, or `--auto` from a SessionStart hook.

Sync an existing project with the latest workflow setup — hooks, GitHub Action, and DB sync template.

Create, destroy, list or sync git worktrees for any project — Herd link+TLS, .env isolation, optional DB clone. Use on `/worktree`.

Fetch PR review comments, implement the fixes, and push.

Fully autonomous SDLC pipeline — plan to PR, zero user intervention. Invoke via /auto-pr.

Capture, list, and promote pre-planning ideas. Use `/backlog` to add, view, promote, complete, or drop ideas before planning.

Verify N PRs, address review feedback, ordered-merge to the staging branch (never main), close linked issues/backlog per merge, then hand off the full test suite. Use on /batch-merge-prs.

Commit staged changes with emoji+type conventions and formatting.

Complete a feature — merge the approved PR, record plan completion, route the homolog→main decision. Use on `/finish` or "finish the feature".

Emergency hotfix deploy — commit, push homolog, PR to main, merge now. Use on `/hotfix <description>`.

Plan and run implementation from a .planning findings dossier — you derive units, waves and model tiers for maximum parallelism, subagents write the code, you verify diffs and commit. Use on /orchestrate [folder].

Post @claude review comment on the current PR and ask what's next. Use on `/pr-review` to trigger the GitHub Actions review workflow.

Create a Pull Request with plan context and base branch detection.

Fast `homolog → main` merge for committed, pushed work. Trigger — `/promote`. Requires out-of-band token minted via `bravros promote unlock` from a separate terminal — Claude cannot mint it.

Safely prune already-merged branches (local + remote) with 7-day tombstone recovery. Manual-only — nothing auto-triggers it. Invoke via `/prune-merged`.