sassy-dog

This skill should be used when the user asks to "assess this repo", "assess it", "audit the codebase", "run a project assessment", "do a full repository health check", "review the whole repo and file issues", "what tech debt do we have", "find risks across the codebase", or wants a comprehensive, evidence-based engineering audit (architecture, security, testing, CI/CD, infra, DX, dependencies, observability) turned into a tracked GitHub Issue backlog. Also triggers for periodic re-assessment, e.g. "a new frontier model dropped, re-assess the repo".

Loop-driven dispatcher: each invocation is one idempotent tick that reconciles in-flight PRs, then tops the pipeline back up to the configured concurrency limit, pulling ONLY from issues marked Ready, respecting dependencies and migration/codegen sequencing, until Ready is empty. Designed to run under "/loop 5m /dispatch-ready" but a single manual invocation is also valid. Use when the user says "let's work ready items", "work ready items", "dispatch ready items", "dispatch ready items in the backlog", "dispatch the backlog", "work the ready queue", "drain it", "drain the backlog", "work through Ready", "keep shipping until Ready is empty", or invokes it via /loop. Reads the current repo's settings from `.claude/sassy-dog/dispatch-ready.md`.

This skill should be used when the user asks to "snapshot the project board", "show the board by status", "pull the backlog with labels", "find stale issues", "which issues shipped but are still open", "find stub issues that need scoping", "file an issue but check for duplicates first", "dedupe before filing", "link this Sentry or feedback item to a GitHub issue", or any task involving GitHub issue/board reads via gh + GraphQL, label taxonomy, stale-issue detection, or gated dedupe-then-file issue creation. Also triggers when a project workflow skill (a generated survey-work or take-it) invokes sassy-dog:github-issues by name.

This skill should be used when the user asks to "set a GitHub secret", "add a GitHub variable", "configure secrets for a workflow", "list GitHub secrets", "fix missing secrets in CI", "set up environment secrets", "add org secrets", "debug empty secret values in GitHub Actions", or any task involving GitHub Actions secrets, variables, or environment configuration. Also triggers when editing .github/workflows/ files that reference secrets.* or vars.* contexts.

Backlog grooming: refine open issues until they are fully dispatchable by a cold worktree sub-agent, then promote them to Ready. The counterpart that feeds dispatch-ready. Use when the user says "groom it", "groom the backlog", "refine the backlog", "scope these issues", "make these dispatchable", "get the backlog ready", "fill it", or asks to move issues to Ready. Writes: issue-body edits, Ready promotion, and epic-split sub-issues only — never deletes, never closes, never dispatches work. Reads the current repo's settings from `.claude/sassy-dog/groom-backlog.md`.

This skill should be used when the user asks to "watch this PR and merge when green", "babysit the PRs", "poll PR checks", "auto-merge the green ones", "enqueue to the merge queue", "why is my PR stuck in the queue", "retry that flaky gh call", "clean up the worktrees after the batch", or any PR-lifecycle mechanics: mergeable checks, check polling, transient-failure retry, merge-queue vs direct merge, serialization of migration/codegen-coupled PRs, post-merge worktree teardown and branch reconcile. Also triggers when a project workflow skill (a generated take-it or send-it) invokes sassy-dog:pr-shepherd by name.

Render the session's end-of-work summary in the fixed recap format — what shipped, what surfaced along the way, which signals deserve a GitHub issue, and the immediate next steps. Use when the user says "recap", "recap the session", "recap the day", "summarize what we did", "summarize the session", "what did we get done", "wrap up the session", "debrief", or asks for a summary of the work just finished. Read-only: it reports and recommends; it never files issues or mutates anything.

Post-shipping git working-state reconciliation mechanics: fast-forward + prune the default branch, sweep `[gone]` and squash-merged local branches, remove stale/detached agent worktrees, sweep orphan `worktree-agent-*` isolation branches, triage orphan stashes by closedByPullRequestsReferences (not bare issue state), sweep untracked-file noise against an allowlist, mop up stale remote branches, and clear orphan claim labels. Use when the user asks "delete the [gone] branches", "why won't this squash-merged branch delete", "triage these orphan stashes", "remove stale agent worktrees", "why are there worktree-agent branches left over", "is this stash safe to drop", or any low-level branch/worktree/stash reconciliation mechanic. Also triggers when a project workflow skill (a generated tidy-repo) invokes sassy-dog:repo-cleanup by name. The repo's own generated tidy-repo owns the top-level post-shipping cleanup request and delegates the mechanics here.

This skill should be used when the user asks to "list TODO/FIXME/HACK markers", "scan tech debt markers", "find skipped tests", "how slow is CI", "CI duration and flake report", "is CI flaky", "is the release pipeline lagging", "mobile release lag", "is TestFlight behind main", or wants a quick scripted signals scan of code-debt markers, CI workflow health (median/p90 duration, flake hints), and release lag — the fast inputs a prioritization pass consumes. Also triggers when a project workflow skill (a generated survey-work) invokes sassy-dog:repo-health by name. For a full multi-agent engineering audit filed as GitHub Issues, use assess-it instead.

Ship a PR end-to-end — worktree audit, freshness gates, pre-flight CI guardrails, template-compliant PR body, commit, push, watch checks, merge, clean up. Use when the user says "send it", "ship it", "open the PR", "create a PR", or asks to merge a branch. Reads the current repo's settings from `.claude/sassy-dog/send-it.md`; run `setup-config` if that file is missing.

This skill should be used when the user asks to "triage Sentry", "pull top unresolved Sentry issues", "what's blowing up in Sentry", "which Sentry errors qualify for a GitHub issue", "cross-reference Sentry with GitHub", "escalate Sentry hits to the backlog", or any task that applies a qualifying gate (unresolved, recent, user/event thresholds) to Sentry issues and optionally files GitHub issues from the survivors. Also triggers when a project workflow skill (a generated survey-work) invokes sassy-dog:sentry-triage by name. For Sentry SDK setup or open-ended natural-language Sentry exploration, defer to the sentry plugin's own skills.

This skill should be used when the user asks to "set up the config for this repo", "set up this repo's sassy-dog config", "configure this repo's workflow skills", "configure survey-work for this repo", "configure send-it here", "set up the workflow config", "bootstrap the sassy-dog config", "refresh this repo's sassy-dog config", "re-sync this repo's workflow config", "update the workflow config", "migrate this repo to config-based workflow skills", "move the workflow skills to config", "adopt the legacy hand-written workflow skills", or "declare the sassy-dog plugin in this repo's settings". Writes and re-syncs a repo's `.claude/sassy-dog/*.md` workflow config plus its `.claude/settings.json` marketplace + plugin declaration, and migrates repos still carrying older generated skills under `.claude/skills/`. Run from inside the target repository.

This skill should be used when the user asks to "set up Dependabot for this repo", "set up dependency automation", "set up dependency updates here", "set up dependabot auto-merge", "set up the dependency workflows", "add dependabot config here", "wire up dependabot auto-merge", "group the dependabot PRs", "stop the dependabot PR flood", "why do dependabot PRs keep failing CI", "fix the bun.lock dependabot problem", "regenerate lockfiles on dependabot PRs", or "standardize dependency updates across our repos". Generates and re-syncs a repo's `.github/dependabot.yml` plus its dependency automation workflows (auto-merge, bun.lock sync, pod lockfile sync) from detected ecosystems. Run from inside the target repository; re-runnable as the stack evolves.

This skill should be used when the user asks to "set up hooks for this repo", "set up the sassydog hooks", "set up per-repo hooks", "set up project hooks here", "set up the format-on-edit hook", "set up a post-edit format hook", "add a formatting hook here", "generate hooks for this repo", "create project hooks for this repo", "make the hooks repo-specific", "wire up format-on-edit for this repo", or "lint on edit in this repo". Creates and re-syncs a consumer repo's stack-specific Claude Code hooks: generated dispatcher scripts under .claude/hooks/sassydog-*.sh plus their .claude/settings.json wiring. Run from inside the target repository; re-runnable as the repo's stack evolves.

This skill should be used when the user asks to "set up this repo", "set this repo up", "bootstrap this repo", "setup repo", "configure this repo", "wire this repo up", "get this repo set up", "onboard this repo", "set up everything for this repo", "new repo setup", or "full repo setup" — the broad whole-repo intent, with no single area named. Orchestrates the three generators strictly in sequence (workflow config, then hooks, then dependency automation), presents one combined plan of every file they would touch, and reports what ran, what was skipped, and why. Delegates all generation and owns none of it. Run from inside the target repository. When the user names one area specifically — config, hooks, or Dependabot — that generator runs directly instead of this one.

Synthesize the full work surface for the current repo — customer pain (Sentry, GitHub bugs, TestFlight feedback), backlog (board or open issues + labels), tech debt (TODO/FIXME, skipped tests), security exposure (code scanning, secret scanning, dependency alerts), dev experience (CI duration/flake), and synthesized "next bet" candidates with no GitHub issue yet. Dedupes across sources, scores within each category, returns a prioritized inline plate. Use when the user says "what's on our plate", "what's on our plate today", "what should we work on", "survey the work", "survey what's on deck", "plate it", "what's next", "what should I prioritize", "give me the plate", "what hurts customers most", "any leaked secrets", "what's our security exposure", or "triage". Reads the current repo's settings from `.claude/sassy-dog/survey-work.md`. Read-only unless that config sets `write_policy: gated`.

Parallel issue-shipping. The user names one or more GitHub issue numbers; dispatch one sub-agent per issue (each in its own git worktree), implement the fix, open a PR with Closes #N, and a coordinator loop polls to auto-merge greens and surface failures. Use when the user says "take #341, #432", "take #N", "take it #N", "go take #N and #M", "pick up #N", "knock out #N", or any variant handing over a list of GitHub issue numbers to ship in parallel. Reads the current repo's settings from `.claude/sassy-dog/take-it.md`.

This skill should be used when the user asks to "check TestFlight feedback", "query TestFlight", "list beta testers", "show TestFlight builds", "get beta feedback", "check App Store Connect", "query app store connect API", "list TestFlight groups", "show build status", or any task involving TestFlight feedback, beta testers, builds, or App Store Connect API queries.

Clean up local + remote git state after a productive day — fast-forward the default branch and prune, inventory stale branches/worktrees, triage orphan stashes, sweep untracked-file noise, remove stale worktrees, delete merged-PR branches local + remote. Use when the user says "clean it", "clean up", "tidy the repo", "tidy up", "tidy it", "clean branches", or asks to remove stale worktrees/branches/stashes. Reads the current repo's settings from `.claude/sassy-dog/tidy-repo.md`; run `setup-config` if that file is missing.

This skill should be used when the user asks "what's behind", "what needs a bump", "what's out of date across our repos", "which products are lagging", "who's on an old version", "portfolio dependency audit", "audit our dependencies across products", "are any products falling behind", "version drift", "which repos have stale actions", "keep the runners updated", or wants a cross-repo currency check spanning EVERY repo in a portfolio rather than one repository. Compares each repo's pinned GitHub Actions, toolchain versions, runner labels, and Dependabot coverage against its peers, then reports which repos are lagging and whether the cause is a missing automation config. Read-only — never edits pins, never opens PRs. For things that are broken or stalled rather than merely old, use whats-on-fire; for single-repo prioritization, use that repository's own survey-work skill.

This skill should be used when the user asks "what's on fire", "what's on fire today", "what's broken", "what's broken across our products", "what's stuck", "what's red", "anything burning", "portfolio status", "portfolio health", "how's the portfolio looking", "which product needs attention", "cross-product status", "any leaked secrets", "security exposure", or wants one cross-product sweep of production failures and stalled work spanning EVERY repo in a GitHub org rather than a single repository. Pulls Sentry issues and cron monitors, org-wide open issues and pull requests, failing workflows, Dependabot exposure, and structural blind spots (products with no error monitoring, no alerting, or no dependency scanning), then ranks them across products and routes each one to the owning repo. Read-only — never files issues, never mutates state. For deep single-repo prioritization, defer to that repository's own survey-work skill.