Skip to content
Skills Library
Management & product · Writing

stories

/stories

This skill should be used when the user runs /rspec:stories or asks to write user stories, acceptance criteria or detailed feature specs for the approved features of the product being reverse-specced. Runs one story-writer subagent per module in parallel.

Repository
rspec ↗
Licence
MIT
Last repository change
11.11.2026

Assessment and security

96/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content31 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 11.11.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

This skill should be used when the user runs /rspec:stories or asks to write user stories, acceptance criteria or detailed feature specs for the approved features of the product being reverse-specced. Runs one story-writer subagent per module in parallel.

Arguments

/stories [module codes, optional; default all modules in scope]
Original author description

This skill should be used when the user runs /rspec:stories or asks to write user stories, acceptance criteria or detailed feature specs for the approved features of the product being reverse-specced. Runs one story-writer subagent per module in parallel.

How it works

SKILL.md has no level-two headings · 12 lines, 829 words

SKILL.md preview
---
name: stories
description: >
  This skill should be used when the user runs /rspec:stories or asks to write user
  stories, acceptance criteria or detailed feature specs for the approved features of the
  product being reverse-specced. Runs one story-writer subagent per module in parallel.
argument-hint: "[module codes, optional; default all modules in scope]"
metadata:
  version: "0.12.0"
---

# /rspec:stories

Requires stage `features-approved` or later. If the stage is `discovered`, stop and ask the user to approve features first.

1. Read `<spec>/features.md` and `scope` in `rspec.json`. The features in scope are the MVP rows, plus the Later rows when `scope` is `full` (the default); never Drop rows. Group them by module code. Put the modules in dependency order, foundations first (e.g. accounts and workspaces before issues, issues before notifications and views); a feature's `depends_on` names only its own module or modules earlier in that order, so parallel writers cannot create cycles between modules; a real need on a later module goes in `also_needs` and is merged after the last round (step 3). If `$ARGUMENTS` lists module codes, restrict to those. On resume, use `pending` from the progress output (MVP first, then Later).
2. Split each module's features into batches of about 10, MVP first. Write in rounds: each round launches one `rspec:story-writer` per module that still has a batch, **all in a single message** (8 at a time if more), so every MVP story is written before any Later one. Give each writer: `track`, product, `lang`, module code and name, its batch's rows verbatim (with priority), the path of its research file, the IDs it may depend on (the features of earlier batches of its own module and of the modules **earlier** in the dependency order, each with its priority), the paths of `<spec>/domain.md` and `<spec>/departures.yaml`, and the absolute path of `${CLAUDE_PLUGIN_ROOT}`.
3. After the last round, run `python3 ${CLAUDE_PLUGIN_ROOT}/scripts/merge_deps.py .`: it moves each `also_needs` into `depends_on` unless it would close a cycle or make an MVP feature need a Later one, and lists what it skipped; a skipped need is a real ordering problem, never dropped silently: split or move one of the two features, or ask the user (autopilot: park it under `## Parked` with both IDs). Then read every story's `## Open questions`, Failure modes and Differences against `domain.md` and against the other stories: add each new shared rule to `domain.md` once, and make the stories that disagree with it agree (or park the conflict); then decide departures (below), then settle unknowns: answer each `Open: unknown:` from the sources where you can (fix the criterion and its tag); move each one still unanswered into `<spec>/unknowns.yaml` as `U-<nnn>` (numbered on from the highest, never reused; `question`, `criteria` from the writer's reply, `probe`, `status: open`, `raised: stories <date>`; under autopilot also park it under `## Parked` in `autopilot-log.md`) and replace its line with `- U-<nnn>: <question>`. Then empty each story's `## Open questions` to `- none`, except its `U-<nnn>` lines, which stay while the unknown is open. An open unknown keeps its criteria at level 0 in `criteria.lock.json` until it is resolved or accepted. For each Differences line, add the criteria it changes to its `departures.yaml` entry's `criteria` list (from the writer's reply), so `criteria.lock.json` links them.
   **Departures found after features approval** (contracts C5; writers never write `departures.yaml`): collect every `Open: departure:` note, every `Differences from the original` line without a `→ D-<nnn>` link, and every Drop row in `features.md` whose reason does not end with one (a spec from before 0.11 has many). A Drop row's entry is `kind: simplification` with its feature in `features`; link it at the end of the row's reason. Ask the user once, AskUserQuestion with up to 4 per question, accept or reject each. Accepted: add a `D-<nnn>` entry (numbered on from the highest, never reused; `kind`, `features`, `criteria`, one-line `summary`, `match: []`, `decided_by: "<user name>, stories <date>"`) and write the story's Differences line ending `→ D-<nnn>`. Rejected: rewrite the affected criteria to the original's behaviour and drop the line. Under autopilot write each as `status: proposed` with no `decided_by`, link it, and park it under `## Parked` in `autopilot-log.md`; a proposed entry is a warning until handoff and an error from then on. After all rounds, run `python3 ${CLAUDE_PLUGIN_ROOT}/scripts/check_spec.py .` and fix any errors (missing stories, criteria breaking contracts C1, missing story sections, leftover placeholders). A spec written before 0.11 first needs `python3 ${CLAUDE_PLUGIN_ROOT}/scripts/migrate_criteria.py .`; each `→ TODO` failure mode it leaves is a warning naming the module to re-run here. Re-run a subagent for a module only if most of its files are broken.
4. The check script already rejects unknown dependencies, an MVP feature depending on a Later one, and cycles. Fix those in the story frontmatter, not in features.md, unless the user agrees to change a priority.
5. Set `stage` to `stories` (never move a later stage back), add a line to `<changelog>`, and reply with: stories written per module, any features whose acceptance criteria rest mostly on `[inferred]` facts, and next step `/rspec:tech`. When the stage was already `tech` or later and criteria changed, the contracts are out of date: say that `/rspec:tech` must run again for the changed stories before the audit or handoff. `check_spec.py` shows it as `tech is stale`, or as `no tech basis` for a spec whose tech ran before 0.11.

Versions

RepositoryScoreLicence
This page · tammai/rspec96MIT
darioabc/massimo100MIT
andriiklymiuk/corgi91MIT

Database snapshot: 13.11.2026 · package version 32cb6a36b81e (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

  • 100write-user-stories
    Write user stories with acceptance criteria. Use when the user says "write user stories", "convert this to stories", "create acceptance criteria", "break this into tickets", "story map", "walking skeleton", "story for this feature", or wants to translate a feature idea into…
  • 100create-user-stories
    Generate user story map with BDD acceptance criteria when the user asks to create user stories, write stories, or break down a feature into stories
  • 95user-stories
    Write INVEST-compliant user stories with Given-When-Then acceptance criteria. Use when writing user stories, creating acceptance criteria, or during /design Step 4.
  • 95memstack-product-user-story-generator
    Use this skill when the user says 'user stories', 'write stories', 'backlog', 'sprint planning', 'acceptance criteria', or needs prioritized stories with Given/When/Then criteria and story point estimates. Do NOT use for full PRDs or detailed feature specs.
  • 100user-story-writer
    Converts feature ideas or rough bullets into user stories with acceptance criteria. Use when asked to write user stories or break down a feature.
  • 90user-story-writing
    Write or refine user stories, scenarios, and acceptance criteria for a software feature. Use when a request must be understandable to product, engineering, design, and QA.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.