Configure Forge

Configure Forge

Uses Workspace Configurator to read the workspace's forge topology and tooling preferences, then generate workspace-level prompts and agents that provide concrete forge operations.

Prerequisites: workspace.config.yaml exists with forge topology and board configuration

Phase 1: Read Configuration

Extract Forge Topology

1. Read workspace.config.yaml
2. Extract the following settings:
   • forge.topology — which providers handle repos, boards, CI, releases
   • forge.tooling.preferred — preferred tool surface (mcp, cli, api)
   • forge.tooling.<provider> — provider-specific tool settings
   • board.* — board field IDs and option IDs
   • adapters.ai-runtimes — which runtimes need generated artifacts
3. Present extracted configuration for confirmation

Output

## Context Anchors

- **Workspace:** <name>
- **Phase:** 1 — Read Configuration

## Forge Topology

| Setting                   | Value   |
| ------------------------- | ------- |
| `forge.topology.repos`    | <value> |
| `forge.topology.boards`   | <value> |
| `forge.topology.ci`       | <value> |
| `forge.topology.releases` | <value> |
| `forge.tooling.preferred` | <value> |

## Board Configuration

| Field          | ID      |
| -------------- | ------- |
| project_id     | <value> |
| status field   | <value> |
| priority field | <value> |

## Next Step

Confirm topology and preferences before generating artifacts.

**Approval Required:** Yes

⛔ CHECKPOINT

STOP. Do not proceed until human explicitly approves:

• Topology settings are correct
• Tooling preferences are appropriate
• Board configuration is complete

Phase 2: Determine Artifacts

Artifact Matrix

Based on topology and tooling preferences, determine what to generate:

Per-Runtime Output

Phase 3: Generate

Create Forge Artifacts

1. Generate forge operation prompts for each declared concern (issues, PRs, board)
2. Generate a forge-ops agent that knows the workspace's specific tooling
3. Include board field IDs and option IDs in generated artifacts
4. Include error handling for authentication and connectivity

Generated Prompt Structure

# Forge: <Operation Name>

> <What this prompt does>

## Configuration

- **Provider:** <from workspace.config.yaml>
- **Tool surface:** <mcp | cli>
- **Board project:** <project_id>

## Operations

### Create Issue

<Provider-specific commands/tool calls for creating an issue>

### Update Board Status

<Provider-specific commands for setting board field values>

### Create PR

<Provider-specific commands for creating a pull request>

Output

## Context Anchors

- **Workspace:** <name>
- **Phase:** 3 — Generate
- **Forge topology:** <repos>/<boards>/<ci>/<releases>

## Proposed Artifacts

| File   | Purpose       |
| ------ | ------------- |
| <path> | <description> |

## Next Step

Review generated artifacts before writing.

**Approval Required:** Yes

⛔ CHECKPOINT

STOP. Do not write files until human explicitly approves:

• Generated artifacts are correct
• Tooling preferences are correctly reflected
• Board field IDs are properly included

Phase 4: Write and Verify

Write Generated Artifacts

1. Write generated files to appropriate locations
2. List what was generated
3. Suggest testing the generated prompts

Output

## Context Anchors

- **Workspace:** <name>
- **Forge topology:** <repos>/<boards>/<ci>/<releases>
- **Tooling preference:** <mcp | cli>

## Generated Artifacts

| File   | Purpose       |
| ------ | ------------- |
| <path> | <description> |

## Next Step

Test the generated prompts by running a forge operation.

**Approval Required:** No

Error Handling