using-matt-workflow
Route non-trivial software delivery, foggy initiatives, recurring workflow design, manual setup or migration, and stakeholder questionnaires in human-gated or…
Skill by OlivierFortier
Chalk Labs
Need help deploying using-matt-workflow?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
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/olivierfortier-using-matt-workflow
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: design, productivity
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. 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/using-matt-workflow
What's inside
- Skill name
using-matt-workflow- Description
- Route non-trivial software delivery, foggy initiatives, recurring workflow design, manual setup or migration, and stakeholder questionnaires in human-gated or explicitly autonomous mode. Use before acting when one branch fits or the user asks which Matt workflow to use.
- License
- MIT
- Instructions
- 97 lines in SKILL.md · view on GitHub
Files (2)
- SKILL.md
- openai.yaml · agents/
SKILL.md (first lines)
# Using Matt Workflow Route before acting. State the chosen route and next gate or autonomous continuation, then use only the skills needed for that route. ## First response Start with exactly these fields: ```text Route: delivery | wayfinder | workflow-design | manual-wizard | questionnaire Experimental capability: <name or none> Next gate: <decision required> Writes: <anticipated artifacts or none> ``` For autonomous mode, write `Next gate: none — autonomous mode explicitly authorized`. Otherwise name the next human decision. Keep repository and external state unchanged until that gate or authorization passes. In Plan Mode, keep all work read-only and present proposed artifact contents in chat. ## Control mode Human-in-the-loop is the default. Activate **autonomous** mode only when the user explicitly asks to run the workflow autonomously or end to end without intermediate approvals. - **human-in-the-loop:** Preserve the design, ticket, experimental-activation, and branch-completion gates below. - **autonomous:** Treat the user's one authorization as approval for scoped, reversible workflow decisions and anticipated writes. Self-approve the design, spec, tickets, and experimental capabilities after reviewing them internally. Autonomous control rules take precedence over component instructions to ask, confirm, or wait on discretionary choices. Make the component's recommended choice, record it, and continue; preserve the blocker rules below. In autonomous mode: 1. Infer low-risk missing details, record the assumptions, and choose the smallest safe route. 2. Create and review specs and tickets internally, then continue without pausing for discretionary approval. 3. Activate batch or other experimental capabilities when their criteria fit; announce their pinned status and continue without pausing. 4. Resolve implementation, review, and test failures and continue until the selected route meets its completion criteria. 5. Finish delivery by leaving the tested, committed feature branch and worktree intact and reporting them. Merge, push, cleanup, generated-wizard execution, and other irreversible actions require authority from the original request. 6. Pause only for unavailable credentials, stakeholder answers, or authority for an irreversible or high-risk external action. State the blocker and exact resume point. ## Route Choose the smallest route that fits: - **delivery** — A feature or concrete engineering change. Use `$matt-workflow:grill-with-docs` when decisions remain. Use `$matt-workflow:prototype`, `$matt-workflow:research`, or `$matt-workflow:handoff` only when they answer a real blocker. Send genuinely small, settled work to `$matt-workflow:implement`; otherwise use `$matt-workflow:to-spec`, then `$matt-workflow:to-tickets`. Enter **Start implementation** before either path writes code. - **wayfinder** — Only for work too large or foggy to hold in one session. Delegate to `$matt-workflow:wayfinder`. For a multi-step human Task ticket, offer the experimental manual-wizard route. Resolve decision tickets, then collapse the map through `$matt-workflow:to-spec`. Direct implementation is reserved for efforts proven genuinely small. In autonomous mode, use **Autonomous Wayfinder** below. - **workflow-design** — A recurring personal or operational activity. Announce and activate experimental `$matt-workflow:loop-me` under the current control mode. If the original request includes implementation, route its decision-complete workflow spec back into delivery or wayfinder. When implementation is outside the original request, complete at the workflow spec. - **manual-wizard** — A human-operated setup, migration, credential, or third-party procedure. Announce and activate experimental `$matt-workflow:wizard` under the current control mode. Generate the Bash script in the OS temporary directory unless the user asks to retain it. Return the path for human execution; automatic execution is forbidden in both modes. - **questionnaire** — The user cannot answer and an identifiable stakeholder can. Announce and activate experimental `$matt-workflow:to-questionnaire` under the current control mode. Record the active issue, decision, or spec as return context; pause and resume that same decision when answers arrive. If multiple routes appear plausible and the choice materially changes the work, choose and record the smallest safe route in autonomous mode; otherwise recommend it and ask one question. ## Load subflow instructions After choosing a route and receiving its current control-mode authorization, read the linked skill files required by that branch before acting: - Delivery: [setup](../setup-matt-pocock-skills/SKILL.md), [grill with docs](../grill-with-docs/SKILL.md), [grilling](../grilling/SKILL.md), [domain modeling](../domain-modeling/SKILL.md), [handoff](../handoff/SKILL.md), [prototype](../prototype/SKILL.md), [research](../research/SKILL.md), [to spec](../to-spec/SKILL.md), [to tickets](../to-tickets/SKILL.md), [implement](../implement/SKILL.md), [TDD](../tdd/SKILL.md), and [code review](../code-review/SKILL.md). - Wayfinding: [wayfinder](../wayfinder/SKILL.md). - Experimental: [questionnaire](../to-questionnaire/SKILL.md), [wizard](../wizard/SKILL.md), and [loop design](../loop-me/SKILL.md). Read these after current control-mode authorization. - Git lifecycle: [worktrees](../using-git-worktrees/SKILL.md) and [finish branch](../finishing-a-development-branch/SKILL.md). Only `using-matt-workflow` is implicit. Compose a route by reading relative references after their gates, then continue without asking the user to invoke each component. ## Interview cadence Default to `$matt-workflow:grilling`, one question at a time. ## Autonomous Wayfinder Replace Wayfinder's session stops with one ticket per fresh agent while preserving one-ticket context boundaries: 1. Chart the map, then give each ready Decision, prototype, task, or research ticket to a fresh agent. Research tickets may run in parallel; process other tickets sequentially. 2. After each resolution, update the map, graduate newly visible fog, and continue until the frontier and fog are empty. 3. Collapse the clear map through `$matt-workflow:to-spec`. If the original request includes implementation, route the completed spec through `$matt-workflow:to-tickets` and delivery; otherwise report the completed spec. ## Start implementation Before either delivery implementation path, delegate to `$matt-workflow:using-git-worktrees` once for the delivery run. Respect an existing isolated worktree, verify its baseline, and run `$matt-workflow:setup-matt-pocock-skills` there first when `docs/agents/issue-tracker.md` is missing. ## Execute tickets After tickets receive current control-mode authorization: 1. Work blockers first. Give each ready ticket to a fresh implementation subagent with only the ticket, approved spec, domain docs, and repository instructions. 2. Require the subagent to use `$matt-workflow:implement`, which drives `$matt-workflow:tdd` and commits its ticket. 3. Run `$matt-workflow:code-review` for specification compliance and code quality. Resolve findings before starting the next ticket. 4. After all tickets, run one whole-spec review and the full test suite. 5. In human-in-the-loop mode, delegate to `$matt-workflow:finishing-a-development-branch` and wait for the user's branch-completion choice. In autonomous mode, leave the tested, committed feature branch and worktree intact and report them. Run ephemeral ticket subagents sequentially in the same feature worktree. Keep durable state in approved specs, issue tickets, and commits. ## Completion criteria - **delivery:** Every approved or self-approved ticket is committed; ticket reviews are resolved; the full test suite and whole-spec review pass; branch-completion choices are presented in human-in-the-loop mode or the ready branch and worktree are reported in autonomous mode. - **wayfinder:** Decision and prototype tickets no longer block a coherent map, and the map has collapsed through `$matt-workflow:to-spec`. - **workflow-design:** `workflows/<slug>.md` and `NOTES.md` contain a decision-complete workflow; the original implementation intent determines whether delivery follows. - **manual-wizard:** `bash -n` accepts the generated script, and the user receives its path and human-run instructions. - **questionnaire:** Every question is answerable by the named stakeholder, the return context names the exact unresolved decision, and the workflow pauses there. `using-matt-workflow` owns route classification. Run each loaded component to its completion criterion, return here at the named gate, and continue the selected route. A direct component call completes only that component.
From OlivierFortier/mattpocock-superpowers, MIT.
More from OlivierFortier/mattpocock-superpowers
- 0.515score
tdd
OlivierFortier
Test-driven red-green-refactor
SkillBFreshness: BClaude Code +9 - 0.515score
to-spec
OlivierFortier
Turn a conversation into a spec
SkillBFreshness: BClaude Code +9 - 0.515score
to-tickets
OlivierFortier
Split a plan into tracer-bullet tickets
SkillBFreshness: BClaude Code +9 - 0.515score
to-questionnaire
OlivierFortier
Front-load questions into a doc for someone to answer
SkillBFreshness: BClaude Code +9 - 0.515score
code-review
OlivierFortier
Review a diff on standards and spec
SkillBFreshness: BClaude Code +9 - 0.515score
grilling
OlivierFortier
Stress-test thinking a round of questions at a time
SkillBFreshness: BClaude Code +9 - 0.515score
wizard
OlivierFortier
Generate an interactive setup wizard
SkillBFreshness: BClaude Code +9 - 0.515score
loop-me
OlivierFortier
Spec the workflows you want to build
SkillBFreshness: BClaude Code +9 - 0.515score
prototype
OlivierFortier
Prototype to answer a design question
SkillBFreshness: BClaude Code +9 - 0.515score
wayfinder
OlivierFortier
Map a large effort as decision tickets
SkillBFreshness: BClaude Code +9 - 0.515score
setup-matt-pocock-skills
OlivierFortier
Configure a repo for the skills
SkillBFreshness: BClaude Code +9 - 0.466score
finishing-a-development-branch
OlivierFortier
Safely finish and integrate a development branch
SkillCFreshness: CClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Aug 6, 2026 (2 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from OlivierFortier
- 0.515score
prototype
OlivierFortier
Prototype to answer a design question
SkillBFreshness: BClaude Code +9 - 0.515score
setup-matt-pocock-skills
OlivierFortier
Configure a repo for the skills
SkillBFreshness: BClaude Code +9 - 0.515score
to-questionnaire
OlivierFortier
Front-load questions into a doc for someone to answer
SkillBFreshness: BClaude Code +9 - 0.515score
loop-me
OlivierFortier
Spec the workflows you want to build
SkillBFreshness: BClaude Code +9 - 0.515score
grilling
OlivierFortier
Stress-test thinking a round of questions at a time
SkillBFreshness: BClaude Code +9 - 0.515score
tdd
OlivierFortier
Test-driven red-green-refactor
SkillBFreshness: BClaude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get using-matt-workflow running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
using-matt-workflow skill. Agent Mart. https://agentmart.dev/skills/olivierfortier-using-matt-workflow (last updated 2026-10-24)