spec-tasks
Skill4 starsChanged 52 days agoUnlicenseView on GitHub
Split a feature's requirements.md into independently-deliverable tasks under specs/NNN_feature_slug/tasks/. Use when the user wants to break down, split, or plan tasks for a spec or feature. Writes one NN_task_slug.md per task ordered by dependency, and maintains a tasks/README.md checklist.
6 other projects use a version of this.
Adopted by
Most recently taken up by
- Buddhilive/buddhi-clitheir own version, first seen Nov 1, 2026
- lbruton/SessionFlowtheir own version, first seen Nov 1, 2026
- xfstudio/skillstheir own version, first seen Oct 13, 2026
- cldanielhernandez/claude-skillstheir own version, first seen Aug 8, 2026
- fernandoxavier02/Pipeline-Orchestratortheir own version, first seen Jun 29, 2026
- MarcosCamara01/sddguardtheir own version, first seen May 24, 2026
Used with
Projects with it also use review.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
SentrySkillreview
Fixture review skill.
Also in 3 projects that use this
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- xfstudio/skills★ 9same name
- Buddhilive/buddhi-cli★ 1same name
- fernandoxavier02/Pipeline-Orchestrator★ 1same name
- lbruton/SessionFlow★ 1same name
- cldanielhernandez/claude-skills★ 0same name
- MarcosCamara01/sddguard★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
- .claude/CLAUDE.mdCLAUDE.md
- buildSkill
- grillingSkill
- new-adrSkill
- spec-buildSkill
- spec-ideateSkill
The file
We split this skill into the rules it gives an agent
5 rules. 0 are common, 1 is distinctive, and 4 are very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
1
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 31
Maintain the checklist: Create/update specs/NNN_feature_slug/tasks/README.md — an ordered checklist of the tasks (- [ ] 01 — {title}) that …
- Very specific: line 34
Format: Run deno fmt on the new/changed files, then report and suggest running spec-build.
- Distinctive: line 74
Keep tasks small enough to review as a self-contained diff.
- Very specific: line 75
Never plan a task around mocking the chain.
- Very specific: line 82
When a bullet's rationale asserts current behaviour — this repo's or a third-party tool's — look it up rather than writing it from memory …
“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
spec-tasks
Split a feature's requirements.md into independently-deliverable, chronologically-ordered task files. This is step 2
of the spec-driven workflow: spec-ideate → spec-tasks → spec-build.
When to use
Invoke when the user wants to break down / split / plan the tasks for an existing spec. Requires a
specs/NNN_feature_slug/requirements.md to already exist (produced by spec-ideate).
Procedure
- Read the requirements. Open
specs/NNN_feature_slug/requirements.md. If the target feature is ambiguous, ask which one. - Decompose into vertical slices. Each task must be deliverable on its own: committable, QA-able, and verifiable independently, mapping back to the acceptance criteria. Avoid tasks that only make sense together.
- Plan the API surface. For every task that gives a fragment a new capability — anything a caller could reasonably
want to do through the APIs, whatever kind of fragment it lands in — decide whether it is reusable beyond this
feature per the
buildskill's API surface convention. If it is, the Fastify route, the FastAPI route, their route tests and the Bruno request belong in that task'sChangesandVerification (QA)— not in a later catch-up task. If it is orchestration, say so in the task file so the decision is visible at build time. - Order by dependency. Sequence tasks so each can be completed and verified before the next. Number them
01_,02_, … (zero-padded two digits). - Write one file per task at
specs/NNN_feature_slug/tasks/NN_task_slug.mdusing the canonical structure below.task_slugis snake_case. - Maintain the checklist. Create/update
specs/NNN_feature_slug/tasks/README.md— an ordered checklist of the tasks (- [ ] 01 — {title}) thatspec-buildticks off as it goes. - Update the index. Set the feature's row in
specs/README.mdtoStatus: Planned. - Format. Run
deno fmton the new/changed files, then report and suggest runningspec-build.
Canonical task file structure
# {NN — task title}
**Status:** Todo | **Feature:** [NNN_feature_slug](../requirements.md)
## Goal
{One deliverable outcome for this task.}
## Depends on
- {earlier task numbers this relies on, or "none"}
## Changes
- {files / modules / fragments to touch, at a high level}
- {API surface: the Fastify + FastAPI routes and Bruno request exposing any new fragment capability — or "none —
<why>", e.g. the change is orchestration only}
## Verification (QA)
- {which commands from the build skill to run — e.g. `node --run test`, `uv run mypy`, `cargo clippy`; whenever the API
surface bullet above is not "none", the co-located route tests and the relevant Bruno suite by name}
- {which integrations the tests exercise for real and which are doubled, per the build skill's testing conventions: the
chain is real against the local validator, AWS SDK clients are mocked}
## Definition of done
- {checklist tying back to the requirements' acceptance criteria}Conventions to honour
- Zero-padded two-digit task numbers; snake_case filenames.
- Reference QA commands by name and defer to the
buildskill for their canonical form — don't duplicate the command list here. - Keep tasks small enough to review as a self-contained diff.
- Never plan a task around mocking the chain. The
buildskill's testing conventions (ADR 013) have blockchain RPC and thesolana_*_interfacefragments exercised against the local validator; only AWS SDK clients are doubled. A QA section that says otherwise is a material preflight finding inspec-build. - Surface new fragment capabilities on both APIs (the
buildskill's API surface convention). Spec 001's task 07 wrote the by-index registration read privately inside the poller, and the interface function plus routes were retrofitted afterwards; planning the exposure into the task avoids that round trip. A requirements file saying "no public API" is about the feature's own surface — it does not exempt capabilities the feature adds to fragments. - When a bullet's rationale asserts current behaviour — this repo's or a third-party tool's — look it up rather than
writing it from memory (
grilling's rule: facts are yours to find, decisions are the user's).spec-buildpreflights these premises before implementing, and a wrong one costs a round trip.
Out of scope
- Writing implementation code or running the tasks (that's
spec-build). - Rewriting
requirements.md— treat it as the stable source of truth.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

