hongyime
- 73 skills
- 0 followers
- 2 days ago last updated
- ▌ Auth · hongyimeConnect Google or Microsoft so Clawd can include your calendar and email in morning briefings.
- ▌
- ▌
- ▌
- ▌ Ingest · hongyimeStatus of the cloud-ingestion pipeline (Gmail / Outlook). Runs automatically at 02:00 SGT.
- ▌
- ▌
- ▌
- ▌ Speech · hongyimeDraft a speech with calibrated word count, tone, and structure. Use when the user types `/speech <topic> [--duration Xm] [--tone formal|casual|persuasive]`.
- ▌ Status · hongyimeShow Clawd runtime health. Use when the user types `/status`, asks "are you up?", "what's your memory count?", "how many facts have you ingested?", "when was the last ingestion?", or wants a quick health check.
- ▌
- ▌ Research · hongyimeGenerate a research report from local knowledge-base facts — strict no-web-search privacy invariant. Delivers .docx download link.
- ▌ Skill Authoring 2 · hongyimeApply standalone conventions for creating, editing, validating, or auditing Prawn Skills in a Git clone. Use for SKILL.md changes and skill library maintenance.
- ▌ Agent Machine Bootstrap · hongyime bundleBootstrap any new, rebuilt or virtual machine into a working coding-agent environment: sizing, base tooling, runtimes, agent CLIs, model credentials, agent config, plugins, MCP servers, skill library, remote access and verification. Use for "set up a new machine", "fresh install", "replicate my setup", "move my agents to another machine", "what do I need on a new box", or reprovisioning after a wipe. OS-agnostic; covers physical, VM and VPS targets, driven remotely or run locally.
- ▌ Opencode Bedrock Config · hongyimeConfigure OpenCode to use AWS Bedrock models, and fix subagents pinned to a dead or unfunded provider. Use for "SigV4 authentication requires AWS credentials", Bedrock provider setup, region pinning for global inference profiles, per-agent model overrides, "no credits remaining" from a subagent, and replicating an OpenCode Bedrock setup onto a new machine.
- ▌ Postgres MCP Onboarding · hongyimeEnable PostgreSQL MCP correctly across providers. USE FOR: local Docker Postgres, Supabase, Railway, Neon, managed Postgres URLs. DO NOT USE FOR: SQL query optimization itself.
- ▌ Virtualbox Guest Access · hongyimeOperate a VirtualBox Linux guest VM from its host over SSH or VBoxManage guestcontrol. Use for guest VM access, "cannot SSH into the VM", host-only networking, guestcontrol command patterns, a VPN killswitch blocking the host subnet, Guest Additions health, and installing tooling inside a guest. Applies to any CLI agent, not one specific harness.
- ▌ Skill Add 2 · hongyimeAdd a skill to a standalone Prawn Skills Git clone from GitHub, a URL, pasted instructions, or a described workflow. Use when asked to add or install a skill.
- ▌ Merge Conflict Resolution · hongyimeResolve git merge/rebase conflicts systematically without losing either side's intent. Use when the user says fix this merge conflict, resolve the rebase, there are conflicts, git is stuck on a conflict, or after pulling/ rebasing/merging when markers appear. Covers reading both sides' intent, semantic (not just textual) resolution, and verification before continuing.
- ▌ Skill Create · hongyimeCreate a new reusable skill in a standalone Prawn Skills Git clone. Use when asked to create a skill, teach an agent a workflow, or make future agents know it.
- ▌ Skill Remove · hongyimePlan and carry out an explicitly requested standalone skill retirement with recoverable backups. Use for uninstalling or archiving an exact named skill.
- ▌ Skill Router 2 · hongyimeFind and load on-demand skills from a Prawn Skills Git clone on any supported OS. Use for unfamiliar tooling, specialized tasks, skill discovery, or requests involving documents, media, cloud platforms, writing, or skill maintenance.
- ▌ Skill Update · hongyimeUpdate or improve existing Prawn Skills in a standalone Git clone using observed workflow evidence. Use when asked to refresh a skill or change its behavior.
- ▌ Skill Cleanup 2 · hongyimeAudit and explain duplicate, broken, default, and on-demand skills in a standalone Prawn Skills installation. Use for messy skills or requests to clean up the library.
- ▌ Architecture Decision Records · hongyime bundleRecord or explain significant architectural decisions with context, alternatives, consequences and proposed/accepted/superseded status. Use for ADRs, record this decision, why did we choose this, or durable design rationale.
- ▌ Fleet Skill Audit · hongyime bundleGeneric, fault-tolerant audit of agent skills discovery and health across all fleet machines. Verifies that agents (OpenCode, Codex, Claude Code, Kiro, Cursor, DeepSeek) discover profile skills on disk with valid frontmatter, intact links, and zero secret leaks without requiring active agent logins. Gracefully skips unreachable hosts. Trigger on "audit fleet skills", "check all machine skills", "verify agent skills across machines".
- ▌ MCP Health Repair · hongyimeTroubleshoot OpenCode MCP failures fast. USE FOR: /status failures, timeout errors, connection closed, broken MCP command args, intermittent MCP flapping. DO NOT USE FOR: feature implementation unrelated to MCP health.
- ▌ Verification Loop · hongyime bundleVerify a completed change with the project's actual build, type, lint, test and diff checks before reporting readiness. Use for verify this change, ready for review, prove this works, or completion checks across any supported harness.
- ▌ Bounded Agent Loop · hongyime bundleRun a user-authorized, bounded repair loop with explicit acceptance commands, isolated ownership, attempt and time limits, failure stop conditions, and a resumable handoff. Use for repeated agent repair work, not routine one-pass edits or unattended scheduling.
- ▌ Homelab Pihole Dns · hongyime bundleDiagnose and plan Pi-hole DNS changes in Docker homelabs with Windows clients and Tailscale. Use for Pi-hole health, blocked domains, broken resolution, DNS routing, v6 configuration and resolver rollout readiness.
- ▌ Codebase Onboarding · hongyime bundleExplain an unfamiliar repository using evidenced architecture, entry points, request or job flow, test commands and conventions. Use for onboard me, understand this repo, where do I start, or preparing a codebase orientation.
- ▌ Competitive Upgrade · hongyimeCompare the current repo, app, or product against real live companies, competitors, SaaS tools, and best-in-class examples using current web research, then turn findings into repo-specific upgrade ideas, specs, tickets, or plans. Use when the user says compare this project with competitors, learn from real companies, what features are missing, upgrade this repo based on market research, benchmark against live products, product capability gap, make this more like the best tools, or find what we should build next. Do NOT use for pure marketing teardowns (competitor-teardown) or ideation without research (game-changing-features).
- ▌ Cross Harness State · hongyimeMaintain MOLT-style persistent agent state in a repo's .agents/ folder so any harness (Codex, Claude Code, Cursor, OpenCode) can resume work with zero ambiguity. Use when starting or ending significant work in a repo, when the user says update the state, save progress, journal this, resume from state, what were we doing, or when switching agents/machines mid-task and continuity must survive.
- ▌ Repo Standardization · hongyimeStandardize repositories to a common baseline: repo tiers, Apache-2.0 LICENSE and NOTICE files, GitHub description/homepage/topics metadata, README identity block, secret/identity scanning, compliance reports, and new-repo scaffolding. Use when the user says standardize this repo, make repos compliant, add license and topics, audit my repos, scaffold a new repo, SHELL check, or bring this repo to house standard.
- ▌ Agent Account Isolation · hongyimeRun two or more accounts of the same coding-agent CLI on one machine without cross-contamination. Use for "it used my other account", "wrong account", separating work and personal agent identities, WSL versus host isolation, per-account credentials, or proving which account a CLI is actually using. Covers browser-login CLIs and cloud-credential CLIs, which fail differently.
- ▌ Skill Cleanup · hongyimeAudit, prune, deduplicate, archive, or reorganize the local AI-agent skill library. Use when the user says clean up skills, too many skills, prune installed roots, remove unused skills, find redundant skills, or reduce skill context. Produces a report before archiving canonical skills.
- ▌ Python Testing · hongyime bundleDesign and repair Python tests using the project's unittest or pytest setup. Use for Python fixtures, parametrization, mocks, async tests, boundary cases, regression tests and test isolation. Preserve coursework and framework rules.
- ▌ Webapp Testing · hongyime bundleVerify web applications in a real browser with Playwright: user flows, layout, screenshots, logs, CI artifacts and flaky-test diagnosis. Use for browser testing and end-to-end flows; use react-testing for isolated component tests.
- ▌ Claude Code CLI · hongyimeDelegate tasks to Claude Code CLI in a hidden background process. Use when the user asks to use Claude Code, or when a task needs multi-file reasoning, complex refactoring, or Claude-quality analysis. Spawns claude in --print mode with no visible terminal — output tee'd to $env:TEMP\hermes-claude-out.txt.
- ▌ Codebase Design · hongyimeDesign how a change should be organized in the codebase: module boundaries, dependency direction, file layout, and seams for testing. Use when the user says where should this go, how should we structure this, design the architecture for this feature, split these modules, or when a repo/feature is growing tangled and needs structural decisions before more code lands.
- ▌ Domain Modeling · hongyimeExtract and refine the conceptual model of a problem space: entities, relationships, invariants, and ubiquitous language, then map it to code. Use when the user says model this domain, what are the core entities, design the data model, untangle these concepts, or when code fights the business logic because concepts are muddled (god objects, boolean-flag entities, duplicated meanings).
- ▌ Postplan Upload · hongyimeUse this when generating a substantial execution plan, architecture spec, refactoring blueprint, implementation plan, visual draft report, PR review summary, or when the user asks to upload, publish, share, or make a clickable plan. Use visual-explainer's shared dark/cyan style and diagrams to render standalone HTML, then upload it with `npx postplan upload`.
- ▌ Skill Authoring · hongyimeConventions for creating, editing, pruning, or auditing AI-agent skills on this Windows dotagents setup. Canonical skills live under OneDrive `01 SKILLS/.agents/skills`; PowerShell is preferred; secrets are excluded from scans; use `dotagents install` not `dotagents sync`. Trigger strongly for add a skill, make a skill, edit this skill, clean up skills, audit frontmatter, or any SKILL.md surgery.
- ▌ CLI Agent Router · hongyime bundleChoose and launch an available CLI agent for authorized delegation, independent review or isolated implementation. Use for agent selection, host adapters, bounded parallel work, and tmux/dmux worktree coordination.
- ▌ Project Transfer · hongyimeOne-time, on-demand transfer of proven lessons from a finished project (SOURCE) to another project (TARGET), even across different languages or stacks. Extracts rules, skills, lint/check intent, hooks and workflows from code, dot files, dot folders and agent session logs, then adapts them to the target. Trigger on "transfer lessons", "harvest this project", "reuse what worked from X".
- ▌ Visual Explainer · hongyime bundleExplain ideas, plans, architecture, code changes, comparisons, audits, and workflows with consistent standalone HTML and useful diagrams. Use proactively for most answers where a visual improves understanding, even short answers; trigger for explain how it works, show the flow, compare options, diagram, UML, Mermaid, C4, HTML report, visual explanation, or a clickable plan. Uses a restrained dark theme, one cyan accent, and visual overview, details, next steps. Works with postplan-upload for hosted delivery.
- ▌ Babysit Pr · hongyime bundleMonitor a pull request through review and CI. Use when the user asks to monitor, watch, or babysit a PR.
- ▌ Hill Climb · hongyimeHill climb any scoreable goal — code, research, or UI — by looping implement, judge, fix until a threshold or plateau. Use when the user says hill climb, hill climbing, climb the hill, iterate until it scores, or keep improving until it hits 9, or when they ask the agent to walk them through defining the goal and eval first. Generic, not UI-only.
- ▌ To Tickets · hongyimeBreak an approved spec, plan, or large change into small ordered implementation tickets with dependencies, file scope, and acceptance checks. Use when the user says break this into tickets, turn the spec into tasks, make a work breakdown, create the ticket list, or when work must span multiple sessions or agents.
- ▌ Polish Repo · hongyimePolish a somewhat-done repo with one command: map it, stress-test it, fix it with tests, tighten and secure it, improve the frontend, prove it, optionally climb the weakest area until it scores 9/10, and open a PR. Use when the user says polish this repo, give this project another pass, or review and improve this project. Also works on an early-stage repo with PRD/TDD docs and little code yet — Phases 1–2 give the second opinion and log the gaps in docs/user-stories.md.
- ▌ Opencode CLI · hongyimeDelegate tasks to OpenCode CLI as a hidden background process. Use when the user asks to use OpenCode, or when a task needs Bedrock model access, or when an active OpenCode session already exists. Spawns opencode hidden with output capture in $env:TEMP\hermes-opencode-out.txt.
- ▌ Skill Router · hongyimeRoute user requests to on-demand skills in the OneDrive canonical skill library. Use when a task sounds like it may need a non-installed skill, when the user asks which skill to use, or when work involves media, documents, presentations, writing, cloud platforms, agent setup, skill creation, or unfamiliar specialized tooling.
- ▌ User Stories · hongyimeReview a finished project with the user through user stories: map what a user can actually do, find edge cases, and reconcile the user's mental model with what the code really does, fixing mismatches with their approval. Use when the user says user stories, review this project, is this what I built, or wants a second pass over a somewhat-done project before polishing it.
- ▌ API Contracts · hongyime bundleDesign and verify API contracts shared by consumers and providers. Use for request/response drift, contract-first implementation, compatibility reviews, schema evolution, generated clients, and validating serialized API responses.
- ▌ Bug Diagnosis · hongyimeStructured bug diagnosis producing a written diagnosis document before any fix is attempted. Use when the user reports a bug, asks to investigate why something fails, says diagnose this bug, find the root cause, figure out why X broke, or when a bug is non-obvious, intermittent, or survived previous fix attempts. Pairs with systematic-debugging (the discipline) and backprop (recurrence prevention).
- ▌ Caveman Stats · hongyime bundleShow real token usage and estimated savings for the current session. Requires the external Caveman Claude Code plugin and its registered hooks. Triggers on /caveman-stats. Report unavailable when the runtime integration is absent; never estimate token counts or claim hooks ran without evidence.
- ▌ React Testing · hongyime bundleWrite and repair React component and hook tests with Testing Library and the project's Jest or Vitest runner. Use for form behavior, async state, network mocks, accessible queries, and deciding between component and browser tests.
- ▌ Shell · hongyimeBring a repo to the SHELL house standard: tiers, Apache-2.0 license files, GitHub metadata, README identity, secret scanning, and a compliance report. Use when the user says shell check, standardize this repo, make repos compliant, audit my repos, or bring this repo to house standard.
- ▌ Unslop · hongyimeRemoves AI tells from any writing: superficial -ing phrases, vague attributions, AI vocabulary, and forced rule-of-three structure, preserving meaning and tone. Use when editing docs, PR descriptions, READMEs, or any text that reads AI-generated.
- ▌ Wizard · hongyimeRun a guided, step-by-step interactive workflow with the user: one question or decision at a time, visible progress, resumable state. Use when the user says walk me through this, guide me step by step, set this up with me, help me decide interactively, or for multi-decision setups (configuring a service, choosing options, filling forms, onboarding flows) where dumping one giant form would overwhelm.
- ▌ File Pr · hongyimeFile a concise pull request. Use when the user asks to file, open, or create a PR.
- ▌ To Spec · hongyimeTurn one clear feature, bug fix, or change request into an implementation-ready spec. Use when the user says write a spec for this, turn this into a spec, spec out this feature, make this buildable, or when a discussed change needs precise requirements before coding. Produces or amends SPEC.md-compatible output and hands off to tickets or direct implementation.
- ▌ Backprop · hongyimeBug-to-invariant protocol. When a bug is confirmed or a test fails, trace the root cause, then decide whether a new §I invariant would prevent recurrence and file it into SPEC.md §B via the spec skill. Triggers on test failure, bug report, post-mortem, or after any systematic-debugging session confirms a cause.
- ▌ Takeover · hongyimeTake over existing work. Use when the user asks to take over a PR, branch, or agent thread, or requests a takeover.
- ▌ Full Send · hongyime bundleTakes a change to a merged PR with nobody watching: proves it, reviews it, files it, babysits it, and merges it once the merge gate passes. Use only when the user explicitly says 'full send'.
- ▌ Full Stop · hongyime bundleTrigger this if the user says 'full stop', even as a closing "Full stop." that reads like emphasis.
- ▌ Modernize · hongyime bundleFor getting old prs up to date. Use when asked to rebase a pr or "modernize".
- ▌ Prototype · hongyimeBuild small throwaway proofs that answer one technical question fast. Use when the user says prototype this, spike it, try a quick proof of concept, test if X is feasible, validate the approach, or when a design decision is blocked on unknowns (library behavior, performance, API quirks). Output is a verdict plus evidence, not production code.
- ▌ Skill Add · hongyimeAdd a skill from an upstream GitHub repository, URL, pasted markdown, pasted instructions, or a described tool workflow. Use when the user says add this skill, install this skill, from GitHub, from a link, from pasted text, or make this into a skill.
- ▌ Wayfinder · hongyimeNavigate large, ambiguous initiatives where the destination or path is unclear. Use when the user says figure out how to approach this, this is too vague to spec, help me find a path, map the territory, we need a plan for a big messy project, or when to-spec/wayfinder routing keeps failing because scope itself is unknown. Produces a wayfinding map: knowns, unknowns, options, and a first concrete step.
- ▌ Molt · hongyimeCarry work across agents and machines with MOLT-style persistent state in a repo's .agents/ folder: STATE.md for now, JOURNAL.md for history, handoff docs for deep recovery. Use when the user says save progress, create handoff, resume from state, switch agents, run out of credits, or what were we doing.
- ▌ Spec · hongyimeCreate, amend, or maintain SPEC.md at repo root: goals, contracts, invariants, tasks, and backpropped bugs. Sole mutator of the project spec. Triggers when the user asks to write a spec, start a new spec, add invariants, amend sections, or record a bug. Companion skills: build (executes §T), check (audits drift), backprop (files bugs into §B).
- ▌ Codex · hongyimeRun the installed Codex CLI for user-requested analysis, review, implementation or resuming a specific task, using verified local flags and configured model defaults.