From idea to production through structured artifacts. Route -> Shape -> Build -> Evidence -> Activate.
A Claude Code plugin that provides structured engineering workflow automation for forgeplan users.
Note: Requires forgeplan CLI -- private ForgePlan application, access through project admin. After receiving the binary: place it in PATH and run
forgeplan init -yin your project.Tip: Works great with the global
/forgeskill for quick methodology reference. This plugin adds automation (commands, hooks, agent) on top of it.
/plugin install forgeplan-workflow@ForgePlan-marketplace> /forge-cycle
Step 1: Health check
Active: 5, Draft: 12, Blind spots: 1 (RFC-003 no evidence)
-> Fix blind spot first? [skip/fix]
Step 2: Task
> "Add PDF export for artifacts"
Step 3: Route
forgeplan route -> Depth: Standard, Pipeline: PRD -> RFC
Step 4: Shape
forgeplan new prd "PDF Export" -> PRD-025
Filling: Problem, Goals, FR...
forgeplan validate PRD-025 -> PASS
Step 5: Build
Implementing FR-001..FR-003...
Tests: 12 new, all passing
Step 6: Evidence
forgeplan new evidence "12 tests, audit clean" -> EVID-008
forgeplan link EVID-008 PRD-025
R_eff: 0.00 -> 1.00
Step 7: Activate
forgeplan activate PRD-025 -> active
Step 8: Commit
feat(export): add PDF export for artifacts
Refs: PRD-025, FR-001..003
See also: Process Reference (EN) / (RU) — how /forge-cycle integrates into the complete idea-to-delivery workflow with methodology routing and adversarial review.
| Type | Name | Description |
|---|---|---|
| Command | /forge-cycle |
Full engineering cycle: health check, route, shape, build, evidence, activate, commit |
| Command | /forge-audit |
Multi-expert code audit: logic, architecture, security, tests, performance, docs |
| Agent | forge-advisor |
Background advisor: suggests routing, evidence, and health checks (non-blocking) |
| Hook | Safety hook | Blocks dangerous Bash commands (force push, hard reset, rm -rf /, DROP TABLE) |
| Hook | PRD check | Warns when editing code without an active PRD |
| KB | forgeplan-methodology |
Agentic RAG skill covering the full forgeplan methodology |
| Section | Content |
|---|---|
| Workflow | Route -> Shape -> Build -> Evidence -> Activate pipeline |
| Artifacts | PRD, RFC, ADR, Evidence -- types and templates |
| Depth | Tactical / Standard / Deep / Critical calibration |
| R_eff | Evidence scoring and quality gates |
| Memory | Cross-session memory via CLAUDE.md and Hindsight |
Forgeplan v0.25.0 introduced the 5-rule hint contract — every CLI/MCP output emits one of:
Next: <command>— primary actionOr: <command>— fallbackWait: <condition>— async retryDone.— terminalFix: <command>— error remediation
This plugin now teaches agents to read these markers automatically. Coverage in Forgeplan v0.25.0 binary: 100% (70/70 CLI commands). See methodology skill section 06-output-hints/agent-protocol.md for the full reading protocol.
MIT