Localization for coding agents: an MCP server for your locale files, plus a write guard that asks before an agent hand-edits them. Flutter ARB, iOS .strings/.xcstrings, Android XML, i18next JSON. On Windows, install @stringlane/cli globally first, because the bundled resolver is a POSIX shell script.
Real estate broker toolkit for Qobrix CRM and WhatsApp (WaSender). Manage listings, leads, viewings, follow-ups, send property selections, run resale-reactivation campaigns, import listings from portals, update prices from developer PDFs, and monitor agent lead inactivity.
Transform Software R&D projects, POCs, ADRs, and lab experiments into production-ready roadmaps with prioritized milestones, risk assessment, and clear deliverables. Includes GitHub Next-style graduation criteria and release phases.
Nine researcher personas + four operational skills (researchers-channel, researchers-project-init, researchers-write-study, researchers-review). Final v0.1 plugin in the constellation; companion to all nine sibling plugins. For Claude Code.
Full shipping pipeline: 12 skills + a read-only Opus code-reviewer agent. Audit and review run as one parallel batch, the same reviewer re-checks every fix round (faster, no oscillating scores). Strict gates: code ≥9.5/10, UX 10/10, docs binary pass/fail.
Nine design personas + four operational skills (designers-channel, designers-project-init, designers-write-spec, designers-design-review). Companion to great-minds, great-authors, great-filmmakers, great-publishers, great-marketers, great-engineers. For Claude Code.
Nine counsel personas + four operational skills (counsels-channel, counsels-project-init, counsels-write-memo, counsels-review). Companion to great-minds, great-authors, great-filmmakers, great-publishers, great-marketers, great-engineers, great-designers, great-operators. For Claude Code.
Nine operations personas + four operational skills (operators-channel, operators-project-init, operators-write-plan, operators-ops-review). Companion to great-minds, great-authors, great-filmmakers, great-publishers, great-marketers, great-engineers, great-designers. For Claude Code.
Personal Claude Code skills by Santhosh Gandhi. Includes context-manager, arch-map, frugal-token-usage, foresight-intelligence, review, whybroken, security-audit, post-ai-thinking, product-pressure-test, and usability-heuristics.
An AI executive team (COO, CISO, CTO, CMO, CSO, CFO) plus a Chief of Staff. Bring real business cases to a candid multi-officer deliberation and get an executive brief with minutes.
Feature-and-above solo+AI development methodology. Thirteen skills: intake (human-originated problems and parked ideas before formal triage, plus intake work selection), product-spec (product decomposition + backlog), feature-spec (feature-level product spec + Behavior Registry change-set), project-state (current project dashboard and resume point), behavior-migration (existing-project migration onto Behavior Registry and harness rails), bug-fix (contract-preserving plan for Active Issues and regressions), arch-audit (architecture analysis), arch-redesign (corrective refactor planning), simplify-code (targeted review of unnecessary work), arch-spec (architecture gate or full per-feature design), implementation-plan (behavior-first Superpowers-compatible plan with registry and harness artifact tasks), bootstrap (project foundation: plugin-internal stack baselines, scaffold, audit, CI/gates, empty intake, empty behavior-gate rails, Project State), and ops (last-mile delivery: profile, CD workflow, compose/Dockerfile or static deploy, reverse-proxy wiring, first staging deploy). hi_flow owns the intake/product/project-state/architecture/behavior/planning rails; Superpowers is recommended as the execution layer, not as a replacement for signed hi_flow specs.
Fetches the operating rules from the private claude-audit repo at the start of any session, installs a Stop hook that refuses to end a substantive turn without a plain-English closing block (What I did / Why / Recommendations / TLDR), and adds /wherewasi, which reports what every running Claude Code chat is doing in plain English. Contains no rules itself.
Find, compare, improve, self-check, and plan installation for agent tools across callable Codex tools, connector plugins, local inventory, live GitHub, Hugging Face, skills.sh, npm, Python, Homebrew, and deferred package registries.
The Software directory: what we use, who owns it, what we pay and when that comes round again, and what would break if it went away. Adds and maintains tools, reports renewals by consequence, and backfills from signed agreements, spend, accounting, a named contract folder, the user's own mailbox and bounded Slack context. External systems are read-only, candidates are approved one at a time, and Slack can support Importance only with exact consequence evidence. Needs the setup plugin, which it reads config from and never calls.
PromptWorld — the meta-compiler *World: a WrightMaster CEO + specialist *Wrights compile/compose/publish prompt systems via the nomicon. Installs from Claude Code.
Audit your CRM against MEDDPICC with quoted evidence, measure adoption per rep over time, link it to won and lost deals, and brief every rep each morning. Local-first, read-only, HubSpot and Granola.