specify
Skill0 starsChanged 5 days agoMITView on GitHub
Investigates a problem or an unclear-impact feature and proposes options. Use when the user has a problem to understand before building. Not for scoped work (design).
29 other projects use a version of this.
Adopted by
Most recently taken up by
- shanewwarren/claude-skillstheir own version, first seen Nov 16, 2026
- codeboehm/maximtheir own version, first seen Nov 15, 2026
- ZaxbyHub/opencode-swarmtheir own version, first seen Nov 15, 2026
- danielcherubini/skillstheir own version, first seen Nov 11, 2026
- joeldevz/skynex-pitheir own version, first seen Nov 8, 2026
- Buddhilive/buddhi-clitheir own version, first seen Nov 1, 2026
Used with
Projects with it also use discovery, discover and 7 more.
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.
robzolkosSkilldiscovery
Quick user-centric interview to capture requirements from a time-poor stakeholder.
Also in 3 projects that use this
anombyte93Skilldiscover
Phase 1 of the prd-taskmaster pipeline: brainstorm-driven discovery. Delegates to superpowers:brainstorming in Interactive Mode (one adaptive question at a time), or self-brainstorms in Autonomous Mode when no user is present. Intercepts before the brainstorming chain hands off to writing-plans — this skill owns the exit. Extracts constraints, calibrates scale (Solo / Team / Enterprise), and advances the pipeline to GENERATE.
Also in 4 projects that use this
ibahgatSkillautopilot
Run autonomous multi-stage execution loops until acceptance criteria pass or blockers are reached.
Also in 4 projects that use this
sahinSkillexecute
Also in 3 projects that use this
Microsoft AzureSkillvalidate
Run the full CI validation workflow locally before submitting a PR. Use this as a final check to ensure all Azure DevOps CI pipeline steps will pass.
Also in 3 projects that use this
Microsoft AzureSkillverify
End-to-end app testing pipeline for browser-based web apps on cloud browsers via Azure Playwright Workspaces: verify the live app, author tests for working flows, heal failing tests, then re-run with video and produce one consolidated report. Three phases (app verification, test authoring, test healing) plus a scoped final sanity run; each phase emits a sub-report. Use when asked to "verify my app", "make sure my app works and has good test coverage", "do a full QA pass", "add tests where I'm missing coverage", "fix failing tests and add new ones", "smoke test before deploy", "QA my deployment", or any combined verify-and-test ask. Requires Azure CLI (`az login`) and a Playwright Workspace resource ID — bootstraps configs on first invocation. Browser-based web apps with Playwright only — not for unit/API tests, native mobile/desktop, load/perf, or full accessibility audits.
Also in 10 projects that use this
- +1 more from Azure
Other versions29 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 29
- zaxbysauce/opencode-swarm★ 532same name
- ZaxbyHub/opencode-swarm★ 531same name
- genkovich/sdd★ 187same name
- team-attention/hoyeon★ 173same name
- ArvorCo/Relentless★ 98same name
- jantichy/claude★ 52same name
- mehrad-dm/mastermind★ 23same name
- agent-ix/quoin★ 20same name
- danielcherubini/skills★ 14same name
- stormlightlabs/thunderus★ 4same name
19 more not listed
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 repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- claude-arsenal/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- coverage-gapsSkill
- dep-upgradeSkill
- designSkill
- executionSkill
- explore-ideaSkill
- gate-checkSkill
- githubSkill
- initSkill
- litcriticSkill
- mutmut-reportSkill
- pin-checkSkill
- pypi-releaseSkill
- python-bootstrapSkill
- queue-addSkill
- queue-nextSkill
- queue-statusSkill
- reviewSkill
- session-endSkill
- shipSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 27 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 70
Always include at least one conservative option (minimal change, lowest risk) and one option that addresses the root cause more thoroughly.
- Distinctive: line 79
Gates check: if the host repo has an engineering-standards skill, verify against its gates (objective, scope, impacted services, …
- Very specific: line 81
Decisions log: when the work came from explore-idea, carry its D-N rows here unchanged — superseded rows included
“Common” is measured: kept by at least 6 in 10 of the 27 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
Specify Workflow
CANARY: specify-loaded-2026-06-01-7f9501625b833979
Owns sections 1–4 of status/specification.md: problem statement, affected systems, options, recommendation. Creates the file if it does not exist; appends/updates these sections if it does. Per-task scratch goes in tmp/ (not committed); never in status/.
Steps
Step 1: Understand the problem
Clarify what is actually being asked. Separate symptoms from root causes. When the work
started in the explore-idea skill, take its decisions log as input: each agreed decision is
a constraint here, not a question to reopen. Cite the I-N ids from status/ideas.md the
spec covers in section 1.
- What is happening? — Observable behavior, errors, or gaps
- What should be happening? — Expected behavior or desired outcome
- Since when? — Timeline, triggers, or recent changes that may be related
- Who is affected? — End users, internal teams, other services, clients
- What is the urgency? — Blocking production, degrading performance, or planned improvement?
Output: a clear, one-paragraph problem statement.
Then list success criteria (measurable): each a metric and threshold where one exists (p95_latency_ms <= 200, error_rate < 0.01), because design turns them into per-task Gates and review / ship verify them. Where a condition cannot be a number, state how it will be judged.
Step 2: Identify affected systems
Map which parts of the codebase and infrastructure are involved.
- Primary service(s)/component(s): where the change or fix will happen
- Dependent systems: services, modules, or components that consume from or feed into the primary
- Shared resources: databases, queues, caches, external APIs
- Infrastructure: cloud resources, deployment configs
- Frontends/clients: any UI or API consumer that surfaces the affected functionality
Output: a dependency map listing each system, its role, and whether it needs changes or just validation.
Step 3: Explain impact
For each affected system, assess what happens if the change ships — and what happens if it doesn't.
- Data impact: stored data, integrity, or data flows?
- API impact: public or internal API contracts? Requires versioning?
- Performance impact: latency, throughput, or resource usage?
- User impact: will end users or API clients notice? Will they need action?
- Operational impact: deployment coordination, monitoring changes, runbook updates?
- Risk if nothing changes: cost of inaction?
Output: an impact assessment with severity (Low / Medium / High) per dimension.
Step 4: Propose options
Present 2-3 viable approaches. For each:
- Description: what the approach does in plain language
- Scope: which services and files are touched
- Effort: Small / Medium / Large
- Tradeoffs: pros and cons (technical debt, risk, maintainability)
- Compatibility: backwards compatible? Requires API versioning?
- Dependencies: infrastructure changes, team coordination, client notification?
Always include at least one conservative option (minimal change, lowest risk) and one option that addresses the root cause more thoroughly.
Output: a comparison table of options.
Step 5: Recommend next step
- Recommended option: which and why
- Immediate next action: first thing to do (e.g., "create branch, start with migration in service X")
- Gates check: if the host repo has an engineering-standards skill, verify against its gates (objective, scope, impacted services, compatibility, risk, validation, release readiness)
- Open questions: anything unresolved that needs input before starting
- Decisions log: when the work came from
explore-idea, carry itsD-Nrows here unchanged — superseded rows included
Output: a clear recommendation with action item.
Abbreviation
Abbreviated specify = Steps 1 + 2 (one paragraph each), where the host repo's CLAUDE.md allows it.
Write and check the file
Load references/template.md when creating or updating status/specification.md (sections 1–4), then confirm its structure:
python3 "${CLAUDE_SKILL_DIR}/scripts/validate_spec.py" --input status/specification.mdIt checks shape only: sections 1–4 and the Success criteria block present and filled (5–6 stay pending until design). Exit 1 names what is missing.
Then self-review the spec before the reader is generated, and fix what fails:
- No placeholders — no
TBD,TODO,...or template text left in any section. - No contradictions — the recommendation, the options and the success criteria agree.
- No ambiguous requirement — each criterion reads one way; "fast" or "robust" gets a number or a stated judgment.
- Scope fits one plan — independent subsystems become separate specs.
- Nothing dropped — every decision and constraint from the source conversation, the decisions log included, is still in the spec; diff a rewrite against the conversation, not the previous draft.
Draw any picture as a fenced drawspec block: claude-arsenal:core:init § references/diagrams.md.
Annotatable reader — before the spec is merged or built on
Once the validator passes, generate the reader and give the user the HTML in the same reply, so their notes attach to the section they are about. Proceeding without annotations needs the reviewer to say so.
create_reader.py # in claude-arsenal/scripts/; run via `uv run --with markdown python3`It finds the spec (arsenal/project/*/spec.md or status/specification.md), writes
<repo>-spec-v<N>-<date>-reader.html and …-annotated.md beside it, and prints both paths; commit
them. Load claude-arsenal:core:init § references/annotatable-reader.md when a
returned notes export arrives, when recording approval, or when another document
needs a reader.
For a user-facing problem, ask once at the start whether a screens review (annotated screenshots, same reader) would help; when yes, or when the user asks for screenshots, load that reference's screens section.
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.

