bootstrap
/bootstrap
Initialize the coordinator and implementation-prompt workflow in a fresh or existing repository. Use when the user asks to bootstrap agent-workflow, set up durable ADRs and prompts, or prepare a repository for Linear-first local execution.
- Author
- robzienert ↗
- Repository
- agent-stuff ↗
- Licence
- MIT
- Last repository change
- 10.08.2026
Assessment and security
- Files in this package
- 5
- Scripts
- 1
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 10.08.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
- 1 scripts in this package. Review them before installing, as Claude may execute them.
Scripts: assets/prompts_index.py
What it does and when to use it
Initialize the coordinator and implementation-prompt workflow in a fresh or existing repository. Use when the user asks to bootstrap agent-workflow, set up durable ADRs and prompts, or prepare a repository for Linear-first local execution.
Original author description
Initialize the coordinator and implementation-prompt workflow in a fresh or existing repository. Use when the user asks to bootstrap agent-workflow, set up durable ADRs and prompts, or prepare a repository for Linear-first local execution.
How it works
- Step 1 — Read and confirm scope
- Step 2 — Create the baseline
- Step 3 — Generate and verify
- Step 4 — Continue through the coordinator
- Guardrails
SKILL.md preview
---
name: bootstrap
description: Initialize the coordinator and implementation-prompt workflow in a fresh or existing repository. Use when the user asks to bootstrap agent-workflow, set up durable ADRs and prompts, or prepare a repository for Linear-first local execution.
---
# Bootstrap
Create the durable design and implementation baseline used by `coordinator`, `new-adr`,
`draft-prompt`, `handoff`, and `linear-local-runner`. Linear owns ideas, priority, readiness, and
operational state. The repository owns ADRs, implementation prompts, verification rules, and merge
provenance.
Run this once per repository. Repair missing pieces without replacing existing project content.
## Step 1 — Read and confirm scope
Read existing `README.md`, `AGENTS.md`, and `CLAUDE.md` before asking questions. Ask one batched
round that covers only missing project-shaping information:
1. Project name, vision, and primary toolchain.
2. The core domain terms that must remain precise.
3. The conventions that implementation prompts and agents must follow.
4. The Linear team and project, when Linear-first execution is wanted.
5. One approved implementation unit to specify first, if one exists.
Do not copy a Linear backlog into repository files.
## Step 2 — Create the baseline
Create missing files only:
```text
AGENTS.md
CLAUDE.md
.agents/README.md
.agents/coordinator/SOUL.md
docs/
adrs/
README.md
template.md
0001-record-architecture-decisions.md
0002-file-based-prompt-system.md
prompts/
README.md
template.md
specs/.gitkeep
research/.gitkeep
scripts/tools/prompts_index.py
```
The generated `docs/prompts/INDEX.md` is not written by hand.
### Bundled files
Copy these domain-neutral assets:
- `assets/prompts_index.py` to `scripts/tools/prompts_index.py`
- `assets/adr-template.md` to `docs/adrs/template.md`
- `assets/prompt-template.md` to `docs/prompts/template.md`
- `assets/coordinator-soul-template.md` as the structure for
`.agents/coordinator/SOUL.md`; replace each bracketed field with project-specific language
### Project-specific files
Write these from the project context:
- `AGENTS.md` contains the domain terms, operating rules, verification commands, and pointers to
the workflow documents. If it exists, propose a focused addition. Do not rewrite it.
- ADR-0001 selects append-only MADR files over chat-only decisions or a mutable design log.
- ADR-0002 selects one numbered Markdown prompt per implementation unit, lifecycle frontmatter,
dependency gates, optional `linear_issue` links, isolated worktrees, and merge provenance.
- `docs/adrs/README.md` documents MADR conventions and indexes ADR-0001 and ADR-0002.
- `docs/prompts/README.md` documents the prompt schema, lifecycle, dependency gate, Linear link,
generated index, and generate/check commands.
- `.agents/README.md` explains that Linear owns the work queue and current issue specification. Git
owns durable prompt snapshots, decisions, code, verification, and merge evidence.
- `.agents/coordinator/SOUL.md` uses project terminology while preserving the coordinator doctrine:
design, decide, specify, delegate, review, and keep durable evidence.
- `CLAUDE.md`, when absent, is a short pointer to `AGENTS.md` and `.agents/README.md`.
If the user selected an approved first implementation unit, create or select its Linear issue and
invoke `draft-prompt`. Do not invent additional backlog items.
## Step 3 — Generate and verify
Generate the index even when no prompts exist:
```sh
python3 scripts/tools/prompts_index.py
python3 scripts/tools/prompts_index.py --check
```
## Step 4 — Continue through the coordinator
Report the files created. Tell the user to invoke `coordinator` to inspect Linear state, settle any
required decision, prepare the next prompt, and launch local work.
## Guardrails
- Preserve existing repository content.
- Do not invent build or test commands.
- Do not duplicate Linear issues in a repository inbox.
- Do not hand-edit `docs/prompts/INDEX.md`.
- Do not mark a prompt or Linear issue ready before its dependencies and blocking decisions are clear.
Files in this package (5)
- SKILL.md 4 KB
- assets/adr-template.md 1 KB
- assets/coordinator-soul-template.md 2 KB
- assets/prompt-template.md 2 KB
- assets/prompts_index.py 10 KB
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · robzienert/agent-stuff | 100 | MIT |
| renzrollon/interlock | 100 | MIT |
| vercel-labs/vercel-plugin | 100 | Apache-2.0 |
| bytedance/deer-flow | 100 | MIT |
| Nazhra05/Super-set | 100 | MIT |
| Nazhra05/Super-set | 100 | MIT |
| chaordic-io/discipline | 100 | MIT |
| geekatron/jerry | 100 | Apache-2.0 |
| fullymiddleaged/clawness | 100 | MIT |
Database snapshot: 13.11.2026 · package version ba7bf3115295 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 90bootstrap-project
Set up the playbook in a new or existing repository: AGENTS.md, ADR 0001, the living docs, CI and owner tasks. Run only when the user asks. - 90initialize
Lay down the skillhead project structure (pointer CLAUDE.md, guides, ADRs, launch pack, hook profile, merge gate) in a new or existing repository, then fill the merge gate for its stack. Use when the user asks to initialize, bootstrap or set up a project for skillhead. - 90agent-context-bootstrap
Bootstrap repo-local agent context for Codex and other coding agents. Use when setting up AGENTS.md, docs/agents, issue tracker rules, triage labels, domain docs, ADR locations, validation commands, or skill usage instructions for a repository. - 100symphony-linear-orchestrator
Bootstrap self-improving Symphony + Linear orchestration for a software repository. Use when Codex needs to inspect a repo, review or generate repo-local guidance, shape Linear issues, render a Symphony workflow, launch a high-leverage first execution wave, or run an… - 100initialize-project
Project initialization and setup workflow. Use when starting a new project, setting up AI-assisted development environment, or bootstrapping an existing codebase with Samuel framework structure. - 100draft-prompt
Convert an approved design or Linear issue into a numbered, lifecycle-tracked implementation prompt under docs/prompts/. Use when work is understood well enough to specify for a local agent, or when the user asks to write an implementation prompt.