Skip to content
Agentmart
Submit a tool

go

Go phase.

Skill by epicsagas

IndexedApache-2.0★ 19
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/epicsagas-go

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Go phase. Reads the approved SPEC file, maps Requirements to tasks (plan), executes via TDD (build), and integrates results verifying all Acceptance Criteria.

Use cases: testing

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/go

What's inside

Skill name
go
Description
Go phase. Reads the approved SPEC file, maps Requirements to tasks (plan), executes via TDD (build), and integrates results verifying all Acceptance Criteria.
License
Apache-2.0
Instructions
173 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Go — Build It

**CRITICAL**: Run `HARNESS_DIR=$(epic-harness path)` first. NEVER use `.harness/` in the project directory.

## Execution Modes

This skill has 3 internal modes that run sequentially:

1. **go:plan** — Decompose requirements into ordered, parallelizable tasks
2. **go:build** — Execute each task using TDD (Red→Green→Refactor)
3. **go:integrate** — Merge results, verify all ACs, fix failures

---

## Mode: go:plan (Plan)

Decompose the spec into an execution plan.

### Process

1. **Load the spec:**
   ```bash
   ls -t $HARNESS_DIR/specs/SPEC-*.md | head -1
   ```
   Read the file. Confirm frontmatter has `status: approved`. If not, invoke the **spec** skill first.

2. **Survey the codebase**: Identify relevant files, modules, patterns for each requirement.

3. **Decompose**: Map each Requirement → one or more Tasks. Every task must reference its source requirement:

   ```
   Task 1: [description] — satisfies: R1 — depends on: none — modifies: [file list]
   Task 2: [description] — satisfies: R2 — depends on: Task 1 — modifies: [file list]
   Task 3: [description] — satisfies: R1,R2 (integration) — depends on: Task 1,2 — modifies: [file list]
   ```

4. **Conflict analysis**:
   - Tasks modifying same files → serialize or use worktree isolation
   - Tasks modifying different files → safe to parallelize
   - Don't plan more than 8 tasks — if bigger, split into phases
   - Include "verify integration" as final task if 3+ tasks

5. Show the plan. Get user confirmation (or auto-proceed if in `/orbit`).

6. **Identify risks**: For each task, list potential failure modes:
   ```
   ### Risks
   - {risk}: {mitigation}
   ```

7. **Create feature branch** (standalone invocation only — `/orbit` handles this in Step 3):
   ```bash
   git checkout -b feature/{goal_slug}
   ```

### Execution Order Format

```
- Batch 1 (parallel): Task 1, Task 3
- Batch 2 (sequential): Task 2
```

---

## Mode: go:build (Execute TDD)

For each task, follow this TDD cycle:

### Builder Protocol

1. Read the task description carefully
2. Identify what file(s) to create or modify
3. **Red**: Write the test(s) — confirm they fail
4. **Green**: Write the minimum code to pass
5. **Refactor**: Clean up if needed — no behavior changes
6. Run tests — confirm they pass
7. Report: what was built, what tests pass

### Parallel Execution

Launch sub-agents with these rules:
- `run_in_background: true` for independent tasks (parallel execution)
- `isolation: "worktree"` if and only if parallel tasks modify overlapping files

| Scenario | Parallel? | Same Files? | Isolation? |
|----------|-----------|-------------|------------|
| Task A, B sequential | No | Any | No |
| Task A, B parallel | Yes | No overlap | No |
| Task A, B parallel | Yes | Overlap exists | Yes |

### Subagent Result States

| State | Meaning | Follow-up |
|-------|---------|-----------|
| **DONE** | Task completed, all tests pass | Proceed |
| **DONE_WITH_CONCERNS** | Completed but has warnings | Review. Escalate security/data/breaking issues. |
| **NEEDS_CONTEXT** | Cannot proceed without user input | Prompt user with specific questions |
| **BLOCKED** | Unresolvable error | Try one alternative. If still blocked, report. |

If stuck 3+ times → invoke **agent-introspection** skill.

### Subagent Output Format

```
## Status: [DONE|DONE_WITH_CONCERNS|NEEDS_CONTEXT|BLOCKED]
## Summary: [1-2 sentences]
## Evidence: [test output, file changes]
## Concerns: [only for DONE_WITH_CONCERNS]
## Questions: [only for NEEDS_CONTEXT]
## Blocker: [only for BLOCKED]
```

---

## Mode: go:integrate (Integrate)

After all tasks complete:

1. Categorize each task result by state (DONE/CONCERNS/NEEDS_CONTEXT/BLOCKED)

From epicsagas/epic-harness, Apache-2.0.

More from epicsagas/epic-harness

Signals

Stars
19
Forks
5
Open issues
3
Last commit (this folder)
Jun 25, 2026 (4 months ago)
Last release
Oct 23, 2026 (4 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
Apache-2.0
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from epicsagas

Cite this page

go skill. Agent Mart. https://agentmart.dev/skills/epicsagas-go (last updated 2026-10-24)