Humans

Refresh drift-prone documentation against current code. Reads docs/architecture/freshness-catalog.yml, diffs against the last sweep's upstream/main anchor, regenerates mechanical entries, processes editorial markers, and opens one PR per sweep with a report file.

Autonomous themed tech-debt cleanup. Reads docs/architecture/debt-ledger.yml, rotates to the least-recently-served debt theme, works it for a time budget (default 2h) or until drained, opens one PR, and asks judgment questions inline at the end. Use for daily debt burndown without Peter pointing at a target: grandfathered analyzer rules, architecture-test baselines, obsolete-field reads, cross-section stitching.

Promote QA → production by opening a PR from peterdrier/Humans:main to nobodies-collective/Humans:main with a properly-qualified commit summary. Use when the user says "PR to production", "promote to prod", "PR from origin to upstream", "prod PR", or any variation of pushing batched fork changes upstream. Always use this skill for the two-remote promotion flow even if the user doesn't say "prod" explicitly.

Run multiple Reforge-guided section-refactor lanes in parallel, each in its own worktree/branch off origin/main, with a score-blind adversarial review panel gating every commit, one draft PR per section. Use when the user wants to reduce Reforge surface/internal score across several sections at once via architecturally-correct deletions (dead surface, entity-leak removal, cross-section read-splits) — not by relocation, parameter bags, or accessibility-dodging. Wraps the per-lane process in .codex/skills/humans-refactor. Triggers: 'refactor swarm', 'parallel section refactors', 'run reforge refactor on Users/Tickets/...', 'burn down surface score across sections'. Has an intensity dial (light|standard|deep) and a workflow|solo execution mode to trade token burn against autonomy per unit of output.

Three-axis orchestrator: (1) clean section boundaries — naming, routes, views, ViewModels, DB ownership, cross-section access; (2) internal cohesion — no EF in service layer, caching in service only, proper interfaces, reusable ViewComponents, architecture-test coverage; (3) focused tests — grouped under the section, covering invariants/negatives/triggers, pruned of redundancy. Push-and-bot-review loop. Use when a sizable PR landed, a section shows arch drift, or /pr-review returned a non-trivial violation list. Camps is the gold-standard reference.

Introduce the cross-section read interface boundary (I<Section>ServiceRead) for one section's service per memory/architecture/section-read-write-split.md. Audits the surface, evaluates which methods belong on the read interface, creates a worktree, dispatches a subagent that introduces the interface and migrates non-section callers, opens a PR. Use when the user says 'read split for X', 'split <Section>Service', 'add I<Section>ServiceRead boundary', 'apply the section-read-write-split rule to <Section>', or any variation of carving the cross-section read surface out of a section's full service interface. Reference implementation is Teams (PR 678). Operates on one section per invocation.

Triage application logs, open GitHub issues, pending feedback, in-app Issues, and agent refusals/handoffs. Runs five phases: logs → close shipped issues → feedback → in-app issues → agent FAQ gaps. Use when /whats shows pending feedback, checking app logs, cleaning up shipped issues, or working the in-app Issues backlog.

Autonomous Reforge-guided refactor workflow for the Humans repository. Use when the user wants Codex to target a section such as Shifts, Users, Teams, Tickets, Camps, or Store, create a dedicated worktree/branch from origin/main, run iterative architecture-focused improvements, score each stage with Reforge, gate commits through a score-blind read-only architecture review, append score and verdict details to commit messages after review acceptance, push progress, and open a PR when the loop reaches stasis.

Autonomous tech-debt reduction workflow for the Humans repository. Use when the user wants Codex to improve high-value code quality issues in a dedicated git worktree branched from origin/main, avoid database or migration changes, commit each improvement separately, push progress, and continue until diminishing returns.

Check an externally created PR for coding rule violations against memory/INDEX.md atoms and docs/architecture/code-review-rules.md.

Stryker-driven test cull: delete slop, consolidate redundants, fill mutation-coverage gaps. Score up, test count and lines down. Use when the user says 'trim tests', '/trim-tests', or wants to clean up LLM-generated test bloat. No args picks the worst-scored file; argument scopes to a section, service, or file.

Review a local diff or PR for unnecessary durable surface: new files, public types, interface methods, DTOs/view models, helpers, routes, DI registrations, services, repositories, and dependencies that should reuse existing owners instead.

Run browser-based smoke tests against the running Humans site using the Claude Code Chrome extension. Argument: all | smoke | auth | profile | consent | teams | admin | gdpr | i18n | [url]

Scrap and regenerate the in-flight EF migrations on the current branch as one consolidated migration. Use when migrations have accumulated mid-development cruft (added/removed columns, stacked add-column-to-existing-table fixes, hand-edited SQL that needs to come out, schema changes that should have been one migration but ended up as five), or when the branch has merged main and its migrations are now stuck mid-chain so `dotnet ef migrations remove` is unsafe. Triggers on phrases like 'regen the migrations', 'redo these migrations', 'scrap and regen migrations', 'consolidate the in-flight migrations', 'redo the migration stack', or any time an agent hits the mid-chain situation in `memory/architecture/migration-regen-after-rebase.md`.

Roslyn-powered semantic query CLI for C# solutions. Use when you need to find references, callers, implementations, dependencies, members, or trace call chains in a C# codebase — replaces multi-round grep/read cycles with single precise queries.

Two-phase memory hygiene. Phase 1: scan external Claude Code memory (~/.claude/projects/<slug>/memory/) for durable rules to migrate into the repo and entries already duplicated by repo atoms. Phase 2: audit in-repo memory/, CLAUDE.md, docs/architecture/, and docs/sections/ for dead links, duplication, drift, and bloat.

Audit site navigation — find dead-end features, missing backlinks, and poor discoverability. Produces an improvement list, no code changes.

Autonomous bug-hunt workflow for the Humans repository. Use when the user wants Codex to scan for real bugs, work in a dedicated git worktree branched from origin/main, avoid database or migration changes, commit each fix separately, push progress, and continue until diminishing returns.

Review code changes against linked GitHub issue specs to catch implementation drift — building something plausible but wrong. Checks each acceptance criterion against actual code.