{
  "markdown": "# WIQ Plugin for Claude and Codex\n\nA Claude and Codex plugin that connects [WIQ](https://www.getwiq.ai/) process intelligence with agentic coding assistants, enabling end-to-end workflow automation powered by real process data.\n\n## What is WIQ?\n\nWIQ is an AI-powered process intelligence platform that turns invisible processes into measurable AI impact. It observes how teams actually work across tools and systems, automatically discovers hidden workflows, and generates implementation-ready AI agent specs. Learn more at [getwiq.ai](https://www.getwiq.ai/).\n\n## What This Plugin Does\n\nThis plugin exposes WIQ's data as MCP (Model Context Protocol) tools for Claude and Codex, allowing an agent to:\n\n- **Retrieve automation blueprints** that define step-by-step workflows with escalation paths and tool requirements\n- **Search past work** to see how users handled similar entities and tasks across applications\n- **Learn from prior executions** to see how other agents ran the same blueprint\n- **Execute workflows end-to-end** using available MCP tools (Linear, Slack, GitHub, etc.)\n- **Validate tool readiness** before attempting automation\n\n## Getting Started\n\n### Prerequisites\n\n- [Claude Code](https://docs.anthropic.com/en/docs/claude-code) CLI or [Codex](https://developers.openai.com/codex) installed\n- Access to a WIQ workspace with configured blueprints\n\n### Installation\n\nClone this repository and install the plugin for the host you use.\n\nFor Claude:\n\n```bash\nclaude plugin add /path/to/wiq-claude-plugin\n```\n\nFor Codex, add this repository as a marketplace source, then install the `wiq` plugin from that marketplace:\n\n```bash\ncodex plugin marketplace add /path/to/wiq-claude-plugin\ncodex plugin add wiq@wiq\n```\n\nThe plugin connects to the WIQ API via an MCP server configuration per harness. The WIQ MCP server handles OAuth discovery and registration, so the plugin does not commit a fixed OAuth client ID.\n\nEach config pins its harness in the endpoint URL — `?platform=claude` for Claude, `?platform=openai-codex` for Codex. WIQ reads that to serve the blueprint version optimized for your runtime and to attribute execution logs, so no skill ever has to name a platform. The two configs are separate files because the value differs: Claude auto-discovers `.mcp.json` at the plugin root, while the Codex manifest points at its own file explicitly.\n\n### Plugin Layout\n\n- `wiq/.mcp.json` — MCP server config for Claude (`?platform=claude`)\n- `wiq/.mcp.codex.json` — MCP server config for Codex (`?platform=openai-codex`)\n- `wiq/.claude-plugin/plugin.json` — Claude plugin manifest\n- `wiq/.codex-plugin/plugin.json` — Codex plugin manifest\n- `.claude-plugin/marketplace.json` — Claude marketplace metadata\n- `.agents/plugins/marketplace.json` — Codex marketplace metadata\n\n## Commands\n\nClaude users can invoke these as slash commands. Codex users can invoke the WIQ plugin or the corresponding bundled skill by name.\n\n### `/wiq <Describe your task>`\n\nRuns end-to-end automation inline in your session — one ticket at a time, sequentially. Step execution stays in the main conversation (no subagent handoff) so you can observe progress and interject between steps. The command will:\n\n- Find matching approved blueprints in WIQ\n- Validate that all required MCP tools are connected\n- Fetch matching tickets/inputs and let you choose which to process\n- Execute the blueprint steps for each ticket with escalation path verification\n\nFor parallel processing, run `/wiq` in a separate session — this skill is sequential by design.\n\n### `/wiq-check-tools <Describe your task>`\n\nChecks whether all MCP tools required by a blueprint are available and properly configured before you attempt automation. The command:\n\n- Lists available blueprints matching your description\n- Lets you select one or multiple blueprints to validate\n- Reports missing, unavailable, or disconnected tools\n- Provides setup suggestions for publicly available tools (Linear, Slack, GitHub)\n\n### `/wiq-try <Describe your task>`\n\nA trial run mode that executes a blueprint one ticket at a time with user confirmation after every step. Use this to safely test automation before committing to full runs. The command:\n\n- Follows the same discovery and setup as `/wiq` (blueprint selection, tool validation)\n- Processes tickets sequentially — never in parallel\n- Pauses after every step for user confirmation before proceeding\n- Verifies escalation paths with evidence from actual step outputs\n- Stops immediately on failures or missing tools — no skipping or disclaimers\n- Provides a detailed summary of each ticket's execution result\n\n### `/wiq-blueprint-from-skill <skill name>`\n\nReads the content of another named Claude/Codex skill and creates a WIQ blueprint from it. Use this to turn a skill you've authored into a reusable, refinable blueprint. The command:\n\n- Reads the named skill's content (including any bundled reference files)\n- Maps the skill's procedure into a blueprint: ordered steps, escalation paths, tools, and ticket-matching rules — grounded in what the skill actually says\n- Shows a preview and asks you to confirm before creating anything\n- Calls `create_blueprint` to create a private draft you own, then points you to `update_blueprint` and the WIQ app to refine and submit it for approval\n\n## Key Design Principles\n\n- **Sequential execution** — Blueprint steps run in strict order with no skipping\n- **Evidence-based Escalation Paths** — Conditions are verified against real data, not assumptions\n- **Fail-safe behavior** — Automation stops and escalates on tool unavailability or escalation path violations\n- **Structured reporting** — Clear summaries of actions taken and outcomes\n\n## License\n\n© Copyright 2026 WIQ by Bardeen Inc. All rights reserved.\n",
  "bytes": 5754,
  "sha": "96eeed6e37fee5211ba81649befe1689fa9034c529962083bd6327c682aeebf0",
  "repo_slug": "getwiq/wiq-claude-plugin",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_getwiq_wiq_claude_plugin_wiq_2f9eaed8/readme"
}