Zero-config SDLC onboarding. Detects project environment, harness capabilities, and harness type. Configures TDD mode, generates AGENTS.md, offers optional enforcement hooks, and recommends skills by phase. Single entry point for all harnesses.
Value: Communication -- establish shared understanding of environment, capabilities, and workflow before any work begins.
Single entry point for configuring the SDLC workflow on any harness. Detects the project environment and harness capabilities, configures TDD mode (guided or automated), generates harness-appropriate instruction files, and recommends skills. Never silently installs or modifies anything.
Gather project context silently before asking questions:
!test -f package.json && echo "js" || true
!test -f Cargo.toml && echo "rust" || true
!test -f pyproject.toml && echo "python" || true
!test -f go.mod && echo "go" || true
!test -f mix.exs && echo "elixir" || true
!git rev-parse --is-inside-work-tree 2>/dev/null && echo "git" || true
!ls skills/*/SKILL.md 2>/dev/null | sed 's|skills/||;s|/SKILL.md||' || true
Record: languages detected, git available, skills already installed. If no language files are found, note "none" and ask the user in Step 5.
Probe for delegation primitives (see references/capability-detection.md):
check for Agent tool and skill chaining (always available).
Identify the harness to generate the correct instruction files:
| Signal | Harness |
|--------|---------|
| CLAUDE.md convention, Claude Code tools | Claude Code |
| AGENTS.md convention, Codex tools | Codex |
| .cursor/rules directory | Cursor / Windsurf |
| None of the above | Generic (AGENTS.md only) |
Match the strategy to detected capabilities:
| Capabilities detected | Strategy | Default mode | |----------------------|----------|--------------| | Agent tool | Subagents | Automated | | Neither | Chaining | Automated |
Chaining is always available and runs all TDD phases sequentially in a
single context. Guided mode (/tdd red, /tdd green, etc.) is always
available as a user-controlled alternative. Let the user override.
If Claude Code is detected and the user wants maximum enforcement, offer
to install optional hook templates from skills/tdd/references/hooks/.
Check if the pipeline skill is installed (skills/pipeline/SKILL.md exists). If detected, factory pipeline mode is available and Step 5 includes an additional question.
Question 1: What are you trying to do?
Question 2: How much process structure?
Question 3 (only when pipeline detected): What autonomy level?
Conservative / Standard / Full. Generate .factory/config.yaml with the
chosen level. See references/skill-recommendations.md for skill lists.
Include in generated instruction files that agents should use task lists (TaskCreate, TaskList, or equivalent harness tools) to track work — even for tasks that seem simple. Simple tasks often reveal subtasks, and a visible task list keeps progress transparent and prevents steps from being forgotten.
This convention goes into the AGENTS.md "Conventions" section (see
references/agents-md.md). If the task-management skill is installed, reference it;
otherwise, include the basic convention directly.
Some installed skills ship a references/agents-md-setup.md file that specifies
content to add to AGENTS.md. Scan all installed skills for this file:
!for skill in skills/*/; do
test -f "${skill}references/agents-md-setup.md" && echo "${skill}"
done
For each skill that has one, read its agents-md-setup.md and incorporate the
specified content into AGENTS.md using managed markers for that skill
(<!-- BEGIN MANAGED: skill-name -->). The setup file documents what section
to add or append to and provides the content template.
Size budget: AGENTS.md must stay under 32 KiB total. Per-skill contributions should be concise — a few bullet points or a short section. If the combined contributions would exceed the budget, summarize and point to the skill for details rather than including full content.
Generate harness-appropriate files. For Claude Code, CLAUDE.md uses
@AGENTS.md references instead of symlinks or content embedding to keep
a single source of truth and avoid duplication. See
references/agents-md.md for AGENTS.md best practices (small routing
document, progressive disclosure, managed markers) and
references/harness-files.md for harness-specific generation rules.
If the user selected "Set up team workflow" or "Full", offer the
ensemble-team skill with its three presets (solo-plus, lean, full).
Stage generated files, commit with a descriptive message, and display:
/tdd to start a TDD cycle, or phase-specific commands)Advisory in all modes. Bootstrap detects and recommends; it never silently
modifies. No skill or setting is installed without explicit user
confirmation [H].
[H]: "Override" means the user explicitly
says "yes, install X" — not inferred permission from context.This skill works standalone. It recommends but does not require other skills.
It generates configuration that references the tdd skill for TDD workflow.
Search for places (restaurants, cafes, etc.) via Google Places API proxy on localhost.
Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
Create or update AgentSkills. Use when designing, structuring, or packaging skills with scripts, references, and assets.
Start voice calls via the OpenClaw voice-call plugin.
Notion API for creating and managing pages, databases, and blocks.
Gemini CLI for one-shot Q&A, summaries, and generation.
Category:developer