Install
$ agentstack add skill-gtrabanco-agentic-workflow-plan-feature-interview ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Plan Feature — Interview (internal)
Turn a rough idea into a well-scoped, architecture-respecting feature through a focused interview. Optimizes for catching unknowns before implementation.
When to use
- The
plan-featurerouter calls this when the input is a vague idea (or the user
forces it with --interview) and wants help thinking it through before any code.
Step 0 — Discover the project (always first)
Per the agent guide's Workflow conventions + documentation map, then read what THIS skill needs: the architecture doc, the domain/style docs relevant to the idea's area, the SPEC template, and the roadmap. Goal: ask only what the project does not already answer, and ground every question in the project's real constraints (layering, domain/i18n/SEO/a11y rules, runtime/platform limits, naming conventions).
Process
- Restate understanding. Summarize the idea in one short paragraph — the
problem it solves and for whom. Get a yes/adjust before going deeper.
- Interview to fill the SPEC. Work through the dimensions below in small
batched rounds (use the question tool; group related questions; offer a recommended default each). Skip anything the docs already settle.
- Problem & goal — what changes for the user; success signal.
- Scope — explicitly what is OUT of scope.
- Architecture impact — layers touched, new entities/ports, use-cases,
adapters; honor the dependency rules.
- Data — schema/migrations, source of truth, cache/consistency.
- Cross-cutting — i18n, SEO, a11y, domain rules, security, per the docs
map. Call out which apply.
- UI reference (only when the feature has a UI surface) — is there a
design to build against (design-system components, mockup, Figma)? If the surface is significant and no reference exists, flag that a design pass should precede implementation rather than letting the code improvise.
- Dev scenarios — happy path and failure modes (empty/degraded state,
races, outages) and how to reproduce each locally.
- Acceptance criteria — objective, verifiable conditions for done.
- Dependencies & risks — other features, external services, unknowns.
- Non-goals / future work — deferred to issues, not built early.
- Size — estimate
XS / S / M / L(the SPEC template defines the scale).
XS/S → SPEC-only, single-pass execution; M/L → full artifact set, phased. If L, propose splitting into independently shippable features.
- Be proactive, not passive. Volunteer assumptions and risks the user
didn't mention; flag conflicts with existing architecture or docs; propose the smallest version that delivers the value.
- Traceability. Offer to open a tracking issue (from the feature issue
template); if created, the PR will Closes #n.
- Produce the SPEC. Once dimensions are answered, write the filled SPEC. The
plan-feature router then runs plan-feature-scaffold for the remaining artifacts + roadmap registration.
- Hand off — return exactly (fixed completion report, back to the router):
`` INTERVIEW → SPEC — size: Dimensions resolved: / Open questions: (→ decisions.md) | none Tracking issue: # | declined | n/a → scaffold next (plan-feature-scaffold) ``
Interview discipline
- One topic per round; don't interrogate. 3–6 batched questions max, then
synthesize.
- Always offer a recommended default so the user can move fast.
- Don't ask what the SPEC template, architecture doc, or roadmap already answers.
- Stop once the SPEC can be filled without guesses; capture residual unknowns as
open questions in decisions.md rather than blocking.
Guardrails
- No code, no branch, no dependencies — design only.
- Smallest-shippable mindset; defer extras to issues.
- Respect and cite the project's architecture and style rules in your reasoning.
- Otherwise per the project's Workflow conventions (docs-language).
Relationship to other skills
- Sibling of
plan-feature-from-issue(issue path); theplan-featurerouter
picks between them by input.
plan-feature-scaffold— scaffolds the artifacts once the interview yields a SPEC.execute-phase— executes the phases afterward.
Done when
- The interview resolved every SPEC dimension (or logged it as an open
question), a filled SPEC + artifacts exist, and the user knows the next step.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: gtrabanco
- Source: gtrabanco/agentic-workflow
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.