john-wick
Fully autonomous, resumable, multi-session SDLC pipeline.
Skill by carson-sweet
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/carson-sweet-john-wick
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.
About
Use cases: data
Reviews
Write a reviewNo 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 26, 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/john-wick
What's inside
- Skill name
john-wick- Description
- Fully autonomous, resumable, multi-session SDLC pipeline.
- License
- AGPL-3.0
- Instructions
- 199 lines in SKILL.md · view on GitHub
Agent Skills spec issues
- name: missing
Files (11)
- SKILL.md
- phase-0-bootstrap.md
- phase-1-define.md
- phase-2-plan.md
- phase-3-design.md
- phase-4-implement-prep.md
- phase-5-implement.md
- phase-6-verify.md
- report-format.md
- severity-classifier.md
- state-schema.md
SKILL.md (first lines)
!`bash ${CLAUDE_SKILL_DIR}/../../hooks/read-state.sh session-state`
<preflight-guard>
STOP. Before executing this skill, check: does .sweetclaude/state/sweetclaude.yaml or .sweetclaude/state/phase.yaml exist in the project directory? If NEITHER, do not proceed. Tell the user: "This project is not set up for SweetClaude. Running the pre-flight check now." Then invoke the sweetclaude master skill (Skill tool, skill: "sweetclaude:master") and run its pre-flight. Return here only after the pre-flight passes.
</preflight-guard>
# John Wick Mode
Fully autonomous, resumable SDLC pipeline. Runs from discovery artifacts to merged PR with minimal human involvement. Interactive gates are pre-defined and infrequent.
**Phase contents (read on demand — see Phase Router below):**
- [phase-0-bootstrap.md](phase-0-bootstrap.md) — B1, B2, B3, B4
- [phase-1-define.md](phase-1-define.md) — D1, D2, D3, D4, CK1
- [phase-2-plan.md](phase-2-plan.md) — P1, P2, P3, P4, CK2
- [phase-3-design.md](phase-3-design.md) — DS1, DS2, DS3, DS4, DS5, DS6, DS7, CK3
- [phase-4-implement-prep.md](phase-4-implement-prep.md) — IP1, IP2, IP3, IP4, IP5, IP6
- [phase-5-implement.md](phase-5-implement.md) — IM1, IM2
- [phase-6-verify.md](phase-6-verify.md) — V1, V2, V3, V4, V5
**Cross-cutting reference:**
- [state-schema.md](state-schema.md) — `john-wick.yaml` schema
- [report-format.md](report-format.md) — Test report template
- [severity-classifier.md](severity-classifier.md) — Significant vs not-significant decision rules
- [../process-controls.md](../process-controls.md) — caucus, subagent, WLF, and correction-loop stop rules
---
## Artifact Path Resolution
Before referencing contract, report, or plan paths, check `.sweetclaude/state/session-state.yaml` → `active_work_item.work_dir`:
- If set: substitute `{work_dir}/contracts/` for `.sweetclaude/contracts/`, `{work_dir}/reports/` for `.sweetclaude/reports/`, and `{work_dir}/plans/` for `.sweetclaude/plans/` throughout all phase documents. Pass resolved paths via `--contract` and `--ledger` CLI args to scripts.
- If not set: use the default paths as written in the phase documents.
This applies to all phase files loaded by this skill.
## Entry and Resume
On invocation, read `.sweetclaude/state/john-wick.yaml`.
**If the file does not exist:** Run the Prerequisites Gate, then start at B1.
**If `status: waiting_for_user`:** Present the pending interactive gate from `interactive_gate_pending`. Collect user input. Update `status: active`. Continue from `current_step`.
**If `status: paused` or `status: active`:** Emit one line: "Last completed: {last step in sessions[].steps_completed}. Resuming from {current_step}." Continue from `current_step`. Before continuing: check `interactive_gate_pending.step`. If it is non-null, treat this resume as `waiting_for_user` — re-present the gate before proceeding.
**If `status: complete`:** Tell the user the pipeline is done. Point to the PR URL stored in `created_artifacts` where type=pr.
**If `status: error`:** Show the recorded error, including the last entry in `sessions[].steps_completed` and the value of `current_step` so the user knows where things broke. Then run the Prerequisites Gate (which will re-check prerequisites and enforce check #7). Do not continue until the gate passes.
**State write discipline:** After every step completes, update `john-wick.yaml`: (1) append the just-completed step name to `sessions[-1].steps_completed`, (2) set `current_step` to the next step — before beginning that next step. State always reflects what comes next, never what just finished. This ensures a resume after interruption skips nothing and repeats nothing.
---
## Phase Router
Map `current_step` → phase file → execute step:
| `current_step` prefix | File to read |
|---|---|
| `B*` | `phase-0-bootstrap.md` |
| `D*`, `CK1` | `phase-1-define.md` |
| `P*`, `CK2` | `phase-2-plan.md` |
| `DS*`, `CK3` | `phase-3-design.md` |
| `IP*` | `phase-4-implement-prep.md` |
| `IM*` | `phase-5-implement.md` |
| `V*` | `phase-6-verify.md` |
When entering a phase for the first time, read the entire phase file. When resuming mid-phase, scroll to the `## {step_id}` section.
---
## Prerequisites Gate
Run on first invocation (no `john-wick.yaml`) OR when `status: error` from a previous run. Validate all of the following. On any failure, halt with the specified error message and do not create `john-wick.yaml`.
| # | Check | How | Error if missing |
|---|---|---|---|
| 1 | SweetClaude initialized | `.sweetclaude/state/sweetclaude.yaml` exists | "Run `/sweetclaude:configure init` first." |
| 2 | Personas artifact | `.sweetclaude/` or `docs/` contains a file with "persona" in the name | "Complete product discovery first: `/sweetclaude:plan product-discovery`" |
| 3 | Task analysis with success + failure criteria | A task analysis artifact exists with "success" and "failure" in its content | "Task analysis incomplete. Rerun `/sweetclaude:plan product-discovery`." |
| 4 | Constraints analysis | A constraints artifact exists in `.sweetclaude/` or `docs/` | "Constraints analysis missing." |
| 5 | Dangerously-skip-permissions acknowledged | Present the warning below and require the user to type "I understand" exactly | "John Wick mode requires explicit acknowledgment." |
| 6 | GitHub mode (conditional) | If user selects GitHub mode at B1: `gh auth status` must exit 0 | "GitHub CLI not authenticated. John Wick can help you fix this now." |
| 7 | No active run in error state | If `john-wick.yaml` exists AND this is not an error-recovery invocation: `status` must be `complete` or `paused` | "Previous run is in error state. Inspect `.sweetclaude/state/john-wick.yaml` before restarting." |
| 8 | Compliance context | `.sweetclaude/state/compliance-context.yaml` exists | Note: not a hard block — collect at B4 if absent. Log: "Compliance context missing — will collect at B4." |
**Note on check #7:** When the Prerequisites Gate is invoked from the `status: error` path (error recovery), check #7 is automatically satisfied — the error state IS the active error being recovered. The gate proceeds to re-validate checks 1-5 and allow recovery.
**Acknowledgment warning (prerequisite 5):**
```
⚠ JOHN WICK MODE
This pipeline runs autonomously through the full SDLC — PRD, design,
TDD, implementation, and PR — with minimal human interaction. It will
create branches, write files, run tests, and open a pull request.
Interactive gates are at: B1, B2, B4 (if needed), D4, DS6, V5, and
conditional IM2. Outside these gates, John Wick does not ask permission.
Type "I understand" to proceed.
```
If the user types anything other than "I understand" (case-insensitive): halt.
**After prerequisites pass:** Initialize `john-wick.yaml` with `status: active`, `current_step: B1`, and the discovery artifact paths found during checks. The full schema is in [state-schema.md](state-schema.md).
---
## Interactive Gate Format
When an I-type step is reached:
1. Complete any autonomous work that precedes the gate.
2. Update `john-wick.yaml`: set `status: waiting_for_user`, populate `interactive_gate_pending.step` and `interactive_gate_pending.description`.
3. Commit any new artifacts produced: `chore(john-wick): artifacts at [step] — awaiting gate`.
4. Present the gate:From carson-sweet/sweetclaude, AGPL-3.0.
More from carson-sweet/sweetclaude
- 0.614score
code-tdd
carson-sweet
Internal TDD process skill — invoked by code-feature, code-issue, and code-debt.
SkillAFreshness: A★ 7Claude Code +9 - 0.614score
code-testing
carson-sweet
Run any combination of test suite, mutation testing, security review, and PR pre-check.
SkillAFreshness: A★ 7Claude Code +9 - 0.614score
project-gh-sync-issues
carson-sweet
Bidirectional status sync between local issue files and GitHub Issues.
SkillAFreshness: A★ 7Claude Code +9 - 0.614score
project-gh-import-issues
carson-sweet
Import open GitHub Issues into the local issue store as v4 story files.
SkillAFreshness: A★ 7Claude Code +9 - 0.614score
product-sprint-plan
carson-sweet
Plan a sprint by selecting stories from the backlog, estimating scope, and producing a sprint commitment.
SkillAFreshness: A★ 7Claude Code +9 - 0.611score
john-wick-checkin
carson-sweet
Internal John Wick phase check-in subagent.
SkillAFreshness: A★ 7Claude Code +9 - 0.611score
code-feature
carson-sweet
Build a new feature end-to-end.
SkillAFreshness: A★ 7Claude Code +9 - 0.611score
code-issue
carson-sweet
Implement a GitHub issue end-to-end.
SkillAFreshness: A★ 7Claude Code +9 - 0.611score
code-review
carson-sweet
Code, security, and compliance review.
SkillAFreshness: A★ 7Claude Code +9 - 0.611score
project-backlog-triage
carson-sweet
Structured backlog grooming session.
SkillAFreshness: A★ 7Claude Code +9 - 0.611score
code-verify
carson-sweet
Run verification before claiming work is complete.
SkillAFreshness: A★ 7Claude Code +9 - 0.611score
product-user-tdd-tests
carson-sweet
Transition user stories into Gherkin .feature files.
SkillAFreshness: A★ 7Claude Code +9
Signals
- Stars
- 7
- Forks
- 3
- Open issues
- 4
- Last commit (this folder)
- Oct 21, 2026 (8 days ago)
- Last release
- Oct 24, 2026 (6 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- AGPL-3.0
- Repository
- github.com/carson-sweet/sweetclaude
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from carson-sweet
- 0.614score
product-sprint-plan
carson-sweet
Plan a sprint by selecting stories from the backlog, estimating scope, and producing a sprint commitment.
SkillAFreshness: A★ 7Claude Code +9 - 0.614score
project-gh-sync-issues
carson-sweet
Bidirectional status sync between local issue files and GitHub Issues.
SkillAFreshness: A★ 7Claude Code +9 - 0.614score
code-testing
carson-sweet
Run any combination of test suite, mutation testing, security review, and PR pre-check.
SkillAFreshness: A★ 7Claude Code +9 - 0.614score
code-tdd
carson-sweet
Internal TDD process skill — invoked by code-feature, code-issue, and code-debt.
SkillAFreshness: A★ 7Claude Code +9 - 0.614score
project-gh-import-issues
carson-sweet
Import open GitHub Issues into the local issue store as v4 story files.
SkillAFreshness: A★ 7Claude Code +9 - 0.611score
code-review
carson-sweet
Code, security, and compliance review.
SkillAFreshness: A★ 7Claude Code +9
Cite this page
john-wick skill. Agent Mart. https://agentmart.dev/skills/carson-sweet-john-wick (last updated 2026-10-24)