AGENT SKILL

intent

Captures a change request as a committed intent.md that a non-engineer can author, then promotes an accepted intent into a forge epic and tasks. Use when someone describes a problem or wish for the product ("we need", "users keep", "it should"), when a product owner wants to file work without the task CLI, or when an accepted intent is ready to enter the queue. NOT for vetting whether to build it (/vet-idea) or for project bootstrap (/new-project).

License 许可
MIT
最近更新
2026-10-23
Stars
★ 0
Forks
0

01 前言 Front Matter

name
intent
description
Captures a change request as a committed intent.md that a non-engineer can author, then promotes an accepted intent into a forge epic and tasks. Use when someone describes a problem or wish for the product ("we need", "users keep", "it should"), when a product owner wants to file work without the task CLI, or when an accepted intent is ready to enter the queue. NOT for vetting whether to build it (/vet-idea) or for project bootstrap (/new-project).
allowed-tools
Read, Glob, Grep, Bash, Write, Edit

03 SKILL.md 正文

/intent

Turns a conversation into a version-controlled intent, then turns an accepted intent into queue state. /intent "<problem>" writes intent/<slug>.md; /intent promote intent/<slug>.md [--epic E0X] files an epic and tasks in the forge registry.

Index

File or section Read when…
Capture Writing a new intent from a described problem
Promote An intent has Status: accepted and is ready for the queue
Where it sits Choosing between /vet-idea, /intent, ISA, and /run-epic
_shared/task-triage.md Derived work surfaces while promoting

Artifact chain

The entry point to the artifact chain: intent → PRD/ISA → tasks → PR. The originator does not need the forge CLI, the Algorithm, or a task id. They need to describe a problem well enough that the next person can act on it. This skill gets them there and stops.

Capture: /intent "<problem>"

  1. Brainstorm until concrete. Ask only what the template cannot be filled without: who feels this, what "fixed" looks like, what must not break. Three to five questions is normal. If the description already answers something, do not re-ask.
  2. Write intent/<slug>.md from templates/intent.md. Slug is kebab-case from the title. Status draft, author from git config user.name, date today. Symptoms in Problem, checkable outcome in Proposed outcome, unknowns in Open questions — never invent answers to fill a section.
  3. Show the file and let the author correct it. Their words win over yours.
  4. Commit it as intent: <title> on the current branch. The commit is the audit trail: who asked, when, what was decided. Do not open a PR — acceptance is a status edit by the product owner, not a merge.

Do not design the solution, estimate it, or file tasks here. That is what promote is for, and only after someone with authority has set Status: accepted.

Promote: /intent promote intent/<slug>.md [--epic E0X]

Preconditions: file exists, Status: accepted. Anything else — stop and say which precondition failed. A draft intent is not queue-eligible; a rejected one never is.

An intent is change-sized, not epic-sized. On an existing product it is usually one to a few tasks inside an epic that already exists; only sometimes is it big enough to be its own epic. So the intent links to its tasks, and links to an epic only when it created that epic. A 35-task epic with one intent filed into it is not "described by" that intent, and the skill must never imply it is.

  1. Read the intent and the project ISA (ISA.md) if present, so tasks inherit the project's constraints.

  2. Choose the epic. With --epic, file into it (it must exist) and leave its description alone — the epic is where the work lives, not what the intent is about. Without --epic, the intent is its own epic; create one from the intent title and link it, because here the epic is the intent:

    python3 .claude/scripts/forge/forge.py epic add E0X --name "<intent title>" \
      --description "From intent/<slug>.md"
    
  3. Draft tasks as tracer bullets — the smallest sequence that reaches Proposed outcome. Look first for prefactoring that makes the work easier ("make the change easy, then make the easy change"); it goes first. Then slice:

    • Each task cuts a narrow but COMPLETE path through every layer (schema, API, UI, tests): vertical, NOT a horizontal slice of one layer.
    • A completed task is demoable or verifiable on its own.
    • Each task is sized to fit in a single fresh context window.
    • Each task declares its blocking edges — the tasks that must complete before it can start (--deps). A task with no blockers can start immediately.
    • The exception is a wide refactor (one mechanical change that breaks many call sites at once): sequence it expand → migrate batches → contract, as /refactor defines.

    Each task also names the files or directories it will touch (--scope-dirs / --scope-files) and gets --isa when it is E3 or above. Open questions in the intent that block a task become --deps on a spike task that answers them, not silent assumptions.

  4. Quiz the breakdown; confirm before filing. Present it as a numbered list — per task: Title, Blocked by, What it delivers (the end-to-end behaviour it makes work) — and ask: does the granularity feel right (too coarse / too fine)? Are the blocking edges correct — does each task only depend on tasks that genuinely gate it? Should any be merged or split? Iterate until approved; that approval is the one confirmation. Then file every task through the CLI in dependency order (blockers first, so --deps can name real ids). Never hand-edit docs/tasks/registry.json. After filing, append one line to each task body's ## Notes section: Intent: intent/<slug>.md. That line is how the intent travels — /run-epic reads it into the task's done-criterion and /create-pr checks the diff against it. The task body is the carrier because every downstream skill already reads it.

  5. Mark the intent Status: promoted → T###, T### (E0X) — the task ids are the link, the epic in parentheses is where to find them — and commit both changes together as intent: promote <slug> → T###. The intent stays in intent/ — it is what /create-pr's reviewer and the task ISA are checked against. Tasks filed later for the same intent get appended to that line.

Where it sits

Question Skill
Should we do this at all? /vet-idea
What does the requester want? /intent (this)
How will we know it is done? ISA (/new-feature scaffolds it at E3+)
Do it /run-epic E0X (drains the epic; each task carries its own intent link)

Derived work discovered while promoting follows skills/_shared/task-triage.md — an intent is not a licence to file everything it reminds you of.

<!-- Tracer-bullet slicing and the breakdown quiz adapted from mattpocock/skills to-tickets (MIT) — see THIRD-PARTY-NOTICES.md -->


Project-specific overrides: if SKILL.local.md exists in this directory, read it — it is consumer-owned, survives framework refresh, and wins on conflict. A sidecar that relaxes a gate defined above must state how to prove the gate is wrong in that case. Doctrine: rules/framework-vs-project-root.md.

04 同仓库更多 Skills

rynhardt81
rynhardt81/forgeFlow

wizard

Generates an interactive bash wizard that walks a human through steps only they can perform — opens each URL, says what to click and copy, captures values (hidden entry for secrets), writes .env and GitHub secrets, confirms before irreversible actions. Use when provisioning infrastructure, setting up credentials or CI secrets, walking an unfamiliar third-party dashboard, or running a one-off migration or cutover. Don't invoke this for steps the agent can perform itself.

★ 0
rynhardt81
rynhardt81/forgeFlow

vet-idea

Vets an idea, decision, or plan BEFORE it is built. Runs a single model through five fixed adversarial advisor roles, a peer-review pass, and a chairman synthesis that emits a halt-capable GO / NO-GO / RECONSIDER verdict. Advisory and non-mutating — outputs a report, never writes task state. Use when an idea, feature, or costly hard-to-reverse decision is of uncertain worth and the question is whether to build it at all. NOT FOR settled work (bug fixes, approved features) or for shaping how to build it (the ISA Interview workflow).

★ 0
rynhardt81
rynhardt81/forgeFlow

triage-incident

Production incident triage — from "something is broken for users" to fixed, verified, and learned-from. Pulls the error signal (Sentry/logs/user report), reproduces before touching code, routes the fix through /fix-bug's prod fast path, and closes with a postmortem entry. Use when something is broken for users in production — an outage, data corruption, a Sentry spike, or a crash-rate alert. NOT FOR bugs found in development (use /fix-bug directly) or CI failures (use /diagnose-ci).

★ 0
rynhardt81
rynhardt81/forgeFlow

setup-review-bot

Configures a repo's AI code reviewers — the post-PR mention bot and the pre-push local reviewer — and writes the AGENTS.md the reviewers read. Interviews the user for what the reviewer should weight, then sets git config forge.reviewBot / forge.localReview and populates AGENTS.md at the project root. Use when a repo has no reviewer configured, when review findings are consistently off-target, or when onboarding a new project.

★ 0
rynhardt81
rynhardt81/forgeFlow

security-review

Portable security review for a feature, endpoint, or change before it ships. Orchestrates the existing OWASP review (@security-boss) and secrets scan (security_secrets validator), then adds the pass they miss — an explicit business-logic-flaw walkthrough (IDOR, race/TOCTOU, state-machine bypass, privilege escalation, price/quantity tampering). Advisory and non-mutating — produces a findings report; real findings are filed as tasks. Use when a change touching money, access control, or multi-step state is about to ship, or when the auth looks fine but the flow could be abused. NOT FOR SAST or dependency-CVE scanning (consumer CI owns those).

★ 0
rynhardt81
rynhardt81/forgeFlow

run-epic

Autonomously drains every open task in a single epic. Pulls ready tasks via the forge CLI, executes the work in-context using the matching discipline (fix-bug / new-feature / refactor), files newly-discovered work as follow-up tasks without stopping to ask, opens a PR per completed task via /create-pr, and repeats until the epic has zero ready/in-progress/pending tasks. Use when the user says `/run-epic E##`, "run epic autonomously", "finish all remaining tasks for epic X", or "drive epic X to completion".

★ 0