foundry

v4.7.3

Autonomous build-verify-fix loop for Claude Code. Reads a Forge spec, decomposes it into castings with pre-authored teammate prompts, validates through an 11-dimension F0.9 gate (including F0.7 intent-carrier coverage), builds in parallel waves with server-side evidence re-execution and per-requirement-ID binding, runs up to 8 verification streams (TRACE, FLOW_TRACE, PROVE, RESEARCH_AUDIT, COVERAGE_DIFF, SIGHT, TEST/PROBE, TEST_OBSERVATIONS), grinds defects to zero, then assays with fresh eyes. Brownfield runs consume a flow delta; greenfield and cosmetic runs use the end-state pipeline. Forge plans. Foundry builds.

Claude Code4 Skills1 MCP serverstdio

By AlphaBravo2 GitHub starsUpdated yesterday

Directory evidence

Runtimes: Claude Code
Parsed components: 5 skill or MCP entries
Source updated: Aug 25, 2026
Manifest status: Canonical path parsed

The directory validates manifest shape and source location. It does not execute the plugin or provide a security endorsement. Review the indexing methodology →

Install plugin

Installs for the current user

claude plugin marketplace add IchenDEV/agent-plugin-mkt
claude plugin marketplace update agent-plugin-marketplace
claude plugin install foundry@agent-plugin-marketplace

Paste and run these commands in a terminal with Claude Code. They add and refresh the PluginsMP catalog, then install this plugin.

The installer fetches third-party code from the source repository shown on this page. This directory validates manifest structure and source location, but does not perform a security audit; review the manifest, components, and source before installing.

Get the source manually

$ git clone https://github.com/alphabravo-oss/guild

Clone the source repository, then follow its setup instructions to add the plugin to a compatible client. The plugin root is plugins/foundry/.

Plugin files

plugins/foundry/

├── .claude-plugin/plugin.json
├── skills/prove/SKILL.md
├── skills/sight/SKILL.md
├── skills/temper/SKILL.md
├── skills/trace/SKILL.md
└── .mcp.json

Included Skills4

proveskills/prove/SKILL.md

Relentless spec-to-code verification with fresh eyes. Reads the spec line by line, reads the actual code, and does not stop until every requirement is provably implemented — not just present, but correct. Catches intent gaps, systemic issues, and spec drift that mechanical audits miss.

sightskills/sight/SKILL.md

Deep browser-based UI audit. Clicks every button, fills every form, opens every dropdown, reads every console log — like a scrutinizing QA engineer. Supports headed and headless. Used as foundry's F2 SIGHT verification stream and as a standalone QA tool.

temperskills/temper/SKILL.md

Post-ASSAY micro-domain deep audit. After F4 ASSAY can't find anything else, temper zooms into tiny areas — individual functions, single pages, specific flows — and asks extremely specific questions about whether they actually work. Finds issues that broad audits miss.

traceskills/trace/SKILL.md

Trace every function, every data flow, and simulate user workflows against the spec. Identifies gaps — missing API calls, disconnected data, unwired functions — that tests and builds don't catch.

MCP servers1

foundrystdio

command: uv
args: run --project ${CLAUDE_PLUGIN_ROOT}/mcp-server foundry-mcp --project-root ${CLAUDE_PROJECT_DIR}
env.FOUNDRY_MODEL: ${user_config.model}

MCP configuration uses runtime-provided plugin path placeholders such as ${PLUGIN_ROOT} or ${CLAUDE_PLUGIN_ROOT}. Review the manifest for the runtime-specific expansion rules.

Plugin manifests1

plugins/foundry/.claude-plugin/plugin.json

{
  "name": "foundry",
  "description": "Autonomous build-verify-fix loop for Claude Code. Reads a Forge spec, decomposes it into castings with pre-authored teammate prompts, validates through an 11-dimension F0.9 gate (including F0.7 intent-carrier coverage), builds in parallel waves with server-side evidence re-execution and per-requirement-ID binding, runs up to 8 verification streams (TRACE, FLOW_TRACE, PROVE, RESEARCH_AUDIT, COVERAGE_DIFF, SIGHT, TEST/PROBE, TEST_OBSERVATIONS), grinds defects to zero, then assays with fresh eyes. Brownfield runs consume a flow delta; greenfield and cosmetic runs use the end-state pipeline. Forge plans. Foundry builds.",
  "version": "4.7.3",
  "author": {
    "name": "AlphaBravo"
  },
  "userConfig": {
    "model": {
      "type": "string",
      "title": "Agent model",
      "description": "Model for this plugin's config-steerable agents. Accepted values: opus, sonnet, haiku, fable, inherit. Leave unset to change nothing — each agent's own model pin governs.",
      "required": false
    }
  },
  "mcpServers": {
    "foundry": {
      "command": "uv",
      "args": [
        "run",
        "--project",
        "${CLAUDE_PLUGIN_ROOT}/mcp-server",
        "foundry-mcp",
        "--project-root",
        "${CLAUDE_PROJECT_DIR}"
      ],
      "env": {
        "FOUNDRY_MODEL": "${user_config.model}"
      }
    }
  }
}