The library

observe · Diagnosing latency and errors in production by correlating logs, metrics, and traces

file-browser · Exploring large codebases or archives without opening each file individually

tech-stack-evaluator · Making informed tech choices before committing months to a stack that won't scale

openbrand · Keeping all marketing and product materials aligned with a single brand voice and visual standard

rcode · Multi-week solo or small-team projects where session-to-session coherence and onboarding context are critical.

release-management · Monorepos with 10+ synchronized packages where independent versioning would fracture the release train.

git · Checkpoint commits mid-work where you need one clean verb (commit/push/PR) without the full ship pipeline.

testrail · QA teams that need bidirectional traceability between test code and test-management artifacts.

event-core · High-throughput event processing where per-event context must stay lazy but cached across async boundaries.

gopass · Scripts that need production secrets without exposing plaintext in env files or shell history.

tokenjuice · Agent workflows where terminal noise (git/pnpm/docker output) consumes tokens without adding signal.

pa-callout-cleanup · Code maintenance where obsolete callout metadata clutters readability.

gostack · Microservice teams deploying Go services where the deployment config should not diverge from source control.

ui-design-system · Multi-product design teams that need single-source-of-truth for components, tokens, and interaction specs.

gws · Workflow automation where Google Workspace is the system-of-record (docs as config, sheets as data, drive as file store).

user-story · Agile teams that need structured, testable requirements before a story goes to development.

sniff-couplers · Refactoring efforts to identify high-coupling modules before decoupling them.

jira · Agile teams that need programmatic Jira updates (auto-close resolved issues, bulk field update, cross-project linking).

ux-researcher-designer · Product teams where user research must directly inform and validate design before engineering investment.

usecomputer · Automation of browser-based workflows (data entry, form submission, multi-step processes) without brittle selectors.

journal · Engineering teams that need an audit trail of decisions, blockers, and retrospectives without relying on chat archives.

vendor-management · Quantifying SLA gaps and renewal decisions when a tier-1 vendor has recurring incidents and you need a defensible KEEP|REVIEW|REPLACE verdict.

miudb · Running structured SQL and smoke-testing database connections when you need machine-readable JSON output for scripting or agent tasks.

video-content-strategist · Building a video presence that compounds authority when long-form content already exists and can be atomized into shorts for discovery.

acpx · Delegating coding work to agents over structured protocol when PTY scraping is unreliable and you want persistent multi-turn sessions.

omnimedia · Processing audio/video/images with Gemini or generating images via Codex when you need multimodal analysis with auto-cascade on rate limit.

pac-proxy-agent · Routing HTTP requests through a PAC file when you need dynamic proxy resolution based on destination URL and network rules.

open-design · Mocking up web artifacts in single self-contained HTML when you want instant browser preview with no build step or external resources.

vpe-review · Pressure-testing delivery commitments and hiring plans with throughput-first metrics (DORA, funnel, team structure) when claims lack operational grounding.

concept-forge · Choosing a 1-day spike experiment after brainstorming when you have 4-6 options and need objective ranking by impact, delight, or effort.

wiki-ingest · Ingesting a new source into knowledge wiki when you want automatic cross-reference updates and contradiction detection across existing pages.

plan-audit · Catching gaps and codebase drift in a plan when stakes warrant independent verification beyond same-context red-teaming with clean-context subagent.

wiki-init · Starting a fresh structured LLM Wiki vault when you need three-layer architecture (raw, wiki, index) with schema templates and starter pages.

run-test · Running Ranger automated tests locally against development server when you need fast feedback on test suite before CI push.

plan · Converting a decided design into concrete phased implementation with file changes and success criteria when you have architecture but need execution steps.

wiki-lint · Catching broken links, orphaned pages, and semantic inconsistencies in knowledge base when you want automated health check before sharing wiki.

wiki-log · Viewing recent wiki activity log when you need to track ingests, updates, and semantic checks across pages over time.

qmd · Searching markdown corpus with BM25/vector/LLM reranking when grep is too crude and you need semantic search over notes or design catalogs.

research · Technology selection and architecture decisions requiring rigorous tradeoff analysis.

pi-prompt-template-model · Switching models and loading domain expertise without manual configuration.

write-a-skill · Creating reusable agent capabilities with proper structure and peer validation.

build-vault-universe · Keeping performance-ranked portfolios current without manual list maintenance.

scenario · Pre-implementation risk discovery and comprehensive QA test planning.

x-twitter-growth · Building X audience via algorithm mechanics and competitive content strategy.

scout · Parallel breadth-first navigation before cross-cutting codebase changes.

security · Threat-modeled vulnerability discovery using attacker personas and STRIDE.

wait-for-CI · Blocking deployment workflows until CI resolution without manual polling.

ship · Safe production deployment with automated gate validation.

skill-management · Discovering and managing installed agent capabilities.

global-agent · Running agents that need to maintain shared execution context.