phntm

Plugin phntm from marketplace phntm in phntm7/skills. Commands read from the repo's manifest — 17 days ago.

Skills (27)

• add-mcp-server

  Use when adding, changing, or troubleshooting an MCP server in the Pi coding agent (`pi mcp add`, `~/.pi/agent/mcp.json`, `.pi/mcp.json`, `/mcp`); covers scope, stdio vs HTTP, credentials, OAuth sign-in, and tool exposure.

• agents-md-init

  Use when creating AGENTS.md or CLAUDE.md project instructions; it inspects the repository and produces concise, project-specific guidance.

• agents-md-maintain

  Use when auditing, updating, merging, or synchronizing existing AGENTS.md and CLAUDE.md files; it removes stale guidance and keeps instructions consistent.

• codex

  Use when the user asks to run the Codex CLI (`codex exec`, `codex exec resume`, `codex exec review`) from another agent, for example to get a GPT-family implementation, second opinion, or cross-family code review.

• coding-benchmarks

  Use when comparing current DeepSWE or FrontierCode results or choosing a model and reasoning effort for coding; it provides normalized data and decision guidance.

• context7-docs

  Use when work depends on current, version-specific documentation for a library, SDK, framework, or API; it retrieves relevant docs and examples with Context7 (`ctx7`).

• craft-mcp

  Use when reading, searching, creating, or editing Craft documents, tasks, daily notes, or collections through MCP; it applies safe escaping and linking conventions.

• deep-code-review

  Use when a PR, large feature, refactor, or whole codebase needs a rigorous review; it routes dynamic checks — correctness, architecture, simplicity, naming, tests, security, performance, duplication, AI-slop, adversarial — and returns prioritized, verified findings. Also triggers on "hostile review"

• github-ops

  Use when working with remote GitHub repositories, Actions, PRs, issues, releases, or code search via gh. Use local tools for a checkout, the `web-search-router` skill for open-web research, and a browser for GitHub UI.

• herdr-omp-orchestration

  Use when orchestrating issue implementation through Herdr with OMP implementer and reviewer agents; it manages worktrees, review loops, gated merges, and cleanup.

• herdr-orchestration

  Use when orchestrating multiple issues through Herdr with Claude Code, Codex, or Pi agents; it manages worktrees, cross-family review loops, gated merges, and cleanup.

• himalaya-email

  Use when reading, searching, sending, or managing email with Himalaya v2, including account or backend failures; it provides safe CLI workflows for mail and auth.

• image-annotator

  Use when marking up or comparing existing raster images with callouts, highlights, redactions, zooms, or diff heatmaps; it produces polished image artifacts.

• linkup-search

  Use when Linkup is requested or selected for filtered web search, page extraction, LinkedIn data, batch lookup, or deep research; it provides structured, cost-aware CLI workflows.

• mise-manage

  Use when installing, pinning, upgrading, or troubleshooting tools and runtimes managed by mise; it covers configuration, backends, lockfiles, environments, and supply-chain controls.

• omniroute-ops

  Use when calling or troubleshooting OmniRoute APIs, authentication, quotas, or provider-routing combos; it provides safe workflows for discovery, requests, and failure diagnosis.

• omp-ops

  Use when configuring Oh My Pi (`omp`) settings, model providers, and credentials, running headless or non-interactive sessions, delegating via subagents, messaging peers, supervising jobs/services, resolving internal URLs, or controlling cost-saving handoffs (Prewalk, Plan-yolo).

• parallel-search

  Use when Parallel is requested or selected for excerpt-rich search, page or PDF extraction, entity discovery, enrichment, monitoring, or deep research; it provides cost-aware CLI workflows.

• playwriter

  Use when a task needs automation in the user's logged-in Chrome for UI testing, JS-heavy scraping, screenshots, or frontend debugging; it provides isolated, stateful Playwright workflows.

• prompt-craft

  Write or review LLM prompts, system instructions, agent configurations, and model-specific guidance with clear outcomes and output contracts.

• session-retro

  Retrospect friction from the current agent session — tools, docs, codebase, environment, task wording, or agent missteps — and propose prioritized prose fixes without editing.

• skill-create

  Use when creating or improving a portable Agent Skill, its trigger description, resources, scripts, or metadata; it provides a compatible structure and validation workflow.

• skills-manage

  Use when discovering, installing, removing, updating, running, or scaffolding Agent Skills with the `skills` CLI; it applies correct scope, agent targeting, telemetry, and lockfile workflows.

• surge-cli

  Use when publishing or managing a static site with Surge, including previews, domains, revisions, rollback, or removal; it provides safe CLI deployment workflows.

• tavily-search

  Use when Tavily is requested or selected for news or finance search, page extraction, site mapping or crawling, or cited deep research; it provides cost-aware CLI workflows.

• web-search-router

  Route web search, extraction, crawling, entity lookup, or monitoring across Linkup, Parallel, Tavily, TinyFish, and Firecrawl when multiple engines are available; use the selected engine's skill for execution details, or this skill's inline command for TinyFish and Firecrawl.

• workflow-audit

  Use when analyzing recent coding-agent histories for repeated manual work; it identifies strong candidates for skills, subagents, commands, automations, or instruction updates.

Similar plugins