{
  "markdown": "# MindForge\n\n**An agentic intelligence framework for Claude Code** — orchestrates multi-agent workflows with governance, memory, and autonomous execution. Production-hardened with true parallelism, streaming SDK, and zero-trust security. Install once, get structured AI-driven development with built-in quality gates.\n\n---\n\n## Latest release\n\n**v11.9.5** (2026-08-22) — The release path can no longer strand itself, and the SDK ships.\n11.9.4 published two packages and then failed on the third; because that step sat *before* the\nrelease page and the `stable` dist-tag move, its failure skipped both. Fixed two ways: the steps\nthat finish a release now run ahead of any additive package publish, and a new offline preflight\ngate refuses to reach a publish that the registry will reject. Verified against a worktree at tag\n`v11.9.4` — the exact tree npm rejected — the gate exits 1 and names the file.\n\n**`mindforge-sdk` publishes for the first time since 11.8.0, and for the first time with\nprovenance.** Everything fixed in it across 11.8.1–11.9.4 had reached nobody, including a\n`WebSocketEventStream` reconnect whose unhandled rejection **terminates the caller's process**.\n\nThe previous release, **v11.9.4**, is where the hook gates started actually registering: 11.9.3\nshipped the code and then declined to run it on essentially every project. Measured against the\npublished tarballs — 11.9.3: **11 hook scripts installed, 0 registered**; 11.9.4: **8 registered,\n3 deny-class verified blocking**. That **behaviour change under a patch bump** still applies — the\ninstaller writes `.claude/settings.json` where it previously declined, merging append-only and\nbacking up first. See the BREAKING section in [CHANGELOG.md](./CHANGELOG.md), or\n[RELEASENOTES.md](./RELEASENOTES.md) for human-readable notes.\n\n---\n\n## What is actually enforced\n\nRead this before the install instructions. MindForge ships a large corpus of agent\ninstructions — commands, skills, personas, protocols — and those are advisory: they work by\nbeing in the model's context, and a model can decline them. The parts that would *block* an\naction are hooks. Through 11.9.2 **no channel registered them.** 11.9.3 added the registration code\nbut it declined to run on almost every project, so in practice nothing was enforced there either.\n**As of 11.9.4** both channels register and execute them **on Claude Code**, and nowhere else.\n\n| Capability | Plugin channel | `npx` channel |\n|---|---|---|\n| Slash commands | Yes | Yes |\n| Skills / personas / protocol docs | Yes | Yes |\n| Subagents | Yes | Yes |\n| Audit hash-chain (`bin/verify-audit.js`) | Yes | Yes |\n| **Hooks enforced (can block a tool call)** | **Claude Code only** | **Claude Code + `--local` only** |\n\nWhat that means, measured rather than asserted:\n\n- **The `npx` channel generates the config it never used to ship.** `files[]` has 49 entries and\n  none of them contains `settings`, so no settings file is *published* — instead\n  `bin/installer/hook-registration.js` writes one at install time, merging append-only into any\n  file you already have. Measured on a confined install: **8 hooks registered** into\n  `.claude/settings.json`, of which the installer's own preflight **executed 7 and verified all 3\n  deny-class hooks returning exit 2** before keeping the file. A preflight failure rolls the\n  registration back rather than leaving a config whose commands do not run.\n- **The plugin channel's dispatcher runs.** It previously crashed on every fire —\n  `run-with-flags.js` requires `./lib/hook-flags` and `plugins/mindforge/scripts/lib/` was not\n  copied in. That directory now exists, all **14 path tokens** in\n  `plugins/mindforge/hooks/hooks.json` resolve under the plugin root, and driving the dispatcher by\n  hand returns **exit 2** for `mindforge-block-no-verify` and `mindforge-config-protection`.\n\nStill **not** enforced, deliberately and with a printed reason for each: any runtime other than\nClaude Code (Cursor, Copilot, Gemini/Antigravity, OpenCode), `--global` scope, a self-install\ninside a MindForge checkout, and Windows. Writing a Claude-schema config into `.cursor/` without an\nexecution-verified hook contract would be decorative. Every outcome, including \"not registered\", is\nprinted by the installer and written to `.mindforge/hook-registration.json`.\n\nThree things gate whether a registered hook is *live*, none of them in MindForge's control: the\nharness must be **restarted** (hooks are snapshotted at session start), the project must be\n**trusted** in the harness, and `CLAUDE_PROJECT_DIR` must be set with `node` on the hook PATH —\nif it is not, the commands exit 1 and the gate is simply absent, which is a deliberate trade\nagainst a fail-closed tail that was measured denying benign commands on a fresh clone. See\n*Hooks are installed but nothing is blocked* in `docs/troubleshooting.md`.\n\nSo: on Claude Code, treat MindForge as a policy enforcement point for the 8 registered hooks plus\na tamper-evident audit log; on every other harness, as **governance-by-convention** plus that same\naudit log. Installing it also expands your repository's trust boundary by a large volume of agent\ninstructions — review what you install. The audit chain is verifiable today\n(`node bin/verify-audit.js`).\n\n---\n\n## Install\n\nClaude Code plugin marketplace (no project files written). The plugin's hooks now fire — see\n*What is actually enforced* above for what that does and does not cover.\n\n```bash\n/plugin marketplace add sairam0424/MindForge\n/plugin install mindforge@mindforge\n```\n\nOr the full framework engine via `npx` (writes `.mindforge/` governance, memory, and planning into your project):\n\n```bash\nnpx mindforge-cc@latest --claude --local\n```\n\nAll install channels (global, local, Antigravity, Cursor, Copilot, Gemini CLI, MCP server, combined runtimes, `--minimal`): see [docs/getting-started.md](docs/getting-started.md).\n\n**Upgrading from 11.9.x?** The installer does not overwrite an existing\n`.mindforge/MINDFORGE-SCHEMA.json`, so 11.9.2's armed config validator keeps the older\npermissive schema on a plain upgrade. Run with `--force` if you want the stricter gate. The\ndaily cost cap declared as `[COST_HARD_LIMIT_USD]` in `MINDFORGE.md` was **not enforced** in\n11.9.2; 11.9.3 arms it. An upgrade never rewrites an existing `MINDFORGE.md`, so if yours\npredates the key the cap stays off — add `[COST_HARD_LIMIT_USD] = 25.00` to turn it on.\n\n---\n\n## Verify\n\n```bash\n/mindforge:health              # framework + installation health check\n/mindforge:health --repair     # fix anything the health check flags\n/mindforge:status              # project status snapshot\n/mindforge:next                # auto-discover your first task\n```\n\nFull verification walkthrough: [docs/quick-verify.md](docs/quick-verify.md).\n\n---\n\n## Quick start (new project)\n\n```bash\n/mindforge:init-project\n/mindforge:plan-phase 1\n/mindforge:execute-phase 1\n/mindforge:verify-phase 1\n/mindforge:ship 1\n```\n\n## Quick start (existing codebase)\n\n```bash\n/mindforge:map-codebase\n/mindforge:do I want to plan the next phase\n/mindforge:plan-phase 1\n```\n\n---\n\n## Documentation\n\n- **User Guide:** [docs/user-guide.md](docs/user-guide.md)\n- **Getting started:** [docs/getting-started.md](docs/getting-started.md)\n- **Quick verify:** [docs/quick-verify.md](docs/quick-verify.md)\n- **Troubleshooting:** [docs/troubleshooting.md](docs/troubleshooting.md)\n- **FAQ:** [docs/faq.md](docs/faq.md)\n- **Full tutorial:** [docs/tutorial.md](docs/tutorial.md)\n- **Commands reference (full):** [docs/commands-reference.md](docs/commands-reference.md)\n- **Commands (quick):** [docs/References/commands.md](docs/References/commands.md)\n- **Config reference:** [docs/References/config-reference.md](docs/References/config-reference.md)\n- **SDK:** [docs/References/sdk-api.md](docs/References/sdk-api.md)\n- **Skills:** [docs/References/skills-api.md](docs/References/skills-api.md)\n- **Audit events:** [docs/References/audit-events.md](docs/References/audit-events.md)\n- **Upgrade guide:** [docs/upgrade.md](docs/upgrade.md)\n- **Workflow atlas:** [docs/workflow-atlas.md](docs/workflow-atlas.md)\n- **Security:** [docs/security/SECURITY.md](docs/security/SECURITY.md) (MindForge never stores credentials in files)\n- **Threat model:** [docs/security/threat-model.md](docs/security/threat-model.md)\n- **Architecture:** [docs/architecture/README.md](docs/architecture/README.md)\n- **Contributing:** [docs/contributing/CONTRIBUTING.md](docs/contributing/CONTRIBUTING.md)\n- **Release notes:** [RELEASENOTES.md](RELEASENOTES.md)\n- **CI quickstart:** [docs/ci-quickstart.md](docs/ci-quickstart.md)\n- **Requirements:** [docs/requirements.md](docs/requirements.md)\n- **Release checklist guide:** [docs/release-checklist-guide.md](docs/release-checklist-guide.md)\n- **USPs and features:** [docs/usp-features.md](docs/usp-features.md)\n\n---\n\n## Core workflow\n\n| Command | What it does |\n| :--- | :--- |\n| `/mindforge:init-project` | Requirements interview → creates `PROJECT.md`, `REQUIREMENTS.md`, `STATE.md` |\n| `/mindforge:plan-phase 1 [--ads]` | Discuss scope, research the domain in parallel, create atomic XML task plans |\n| `/mindforge:execute-phase 1` | Wave-based parallel execution, one commit per task, automated verification |\n| `/mindforge:verify-phase 1` | Human acceptance testing, debug agent on failures, UAT sign-off |\n| `/mindforge:ship 1` | Changelog generation, final quality gates, PR creation |\n| `/mindforge:auto --phase 1` | Walk-away autonomous execution with stuck detection and steering |\n\nFull command list: [docs/commands-reference.md](docs/commands-reference.md).\n\n---\n\n## Dynamic Workflow Library\n\n35 pre-built multi-agent workflow scripts that run via Claude Code's `Workflow` tool. Each fans out concurrent agents, synthesizes results, and returns structured output across 5 tiers (Research, Dev, Ops, Intelligence, Beast).\n\n**Discover:** `/mindforge:wf-catalog` or `node bin/mindforge-cli.js workflow list`\n\nFull, verified 35-workflow table by tier: [docs/workflow-atlas.md](docs/workflow-atlas.md).\n\n---\n\n## Execution Modes\n\nMindForge supports multiple interaction models to fit your engineering workflow:\n\n- **In-IDE Orchestration**: Use `/mindforge:agent <persona>` for real-time delegation.\n- **Enterprise Workflows**: Specialized commands like `/mindforge:wf-tdd-sprint` and `/mindforge:plan-phase`.\n- **CLI Automation**: Run `node bin/mindforge-cli.js spawn <persona>` for scripted tasks.\n\n---\n\n## Updates and migrations\n\nRun `/mindforge:update` (add `--apply` to install) — see [docs/upgrade.md](docs/upgrade.md) for the full upgrade guide and fallback steps.\n\n---\n\n## Plugin system (v1.0.0)\n\nPlugins extend MindForge via the `mindforge-plugin-*` namespace.\n\n```\n/mindforge:plugins list\n/mindforge:plugins install mindforge-plugin-<name>\n/mindforge:plugins validate\n```\n\n---\n\n## Token usage profiling\n\n```\n/mindforge:tokens --profile\n```\nSee `.mindforge/production/token-optimiser.md`.\n\n---\n\n## License\n\nMIT © 2026 MindForge Team\n",
  "bytes": 10972,
  "sha": "2f52da42f93931e77db589023065114fe121f9d049af64017dccafa047e14953",
  "repo_slug": "sairam0424/mindforge",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_sairam0424_mindforge_9c13017d/readme"
}