agentic-sdlc

Plugin agentic-sdlc from marketplace agentic-skills in thiagoalmedeiros/agent-skills. Commands read from the repo's manifest — 18 days ago.

Skills (20)

• browser-testing

  Verify browser-facing changes by actually opening a real browser and driving it — navigating, clicking, typing, watching the screen — the way a human tester would, instead of guessing from static code or relying only on structured API output. The mechanism that opens the browser is detected from wha

• code-reviewer

  Multi-axis code review, language- and framework-agnostic — correctness, readability, architecture/SOLID, security, and performance, layered with convention conformance, duplication detection, test quality, and strict comment hygiene (self-documenting code; over-commenting is a readability defect). U

• code-simplification

  Reduces complexity in working code while preserving exact behavior, judged by whether a new teammate would understand the result faster than the original. USE FOR: cleaning up deeply nested logic, long functions, or unclear names after a feature works; consolidating duplicated logic; refactoring cod

• define

  Define the work — turn a task description or Jira ticket into a structured implementation plan at plans/<topic>/plan.md before any code is written, by clarifying intent, stress-testing the approach, then planning. USE FOR: starting a piece of work end to end, "define this", "plan this ticket", turni

• devcontainer-setup

  Set up or review a polyglot-ready .devcontainer environment (Docker Compose + Dockerfile + devcontainer.json) built on the Microsoft devcontainers base images, for any project stack. USE FOR: creating a new .devcontainer for a project, or auditing an existing .devcontainer against the standard. DO N

• docs-sync

  Align documentation content with the actual code changes on a branch — across the four repo-root docs (AGENTS.md, ARCHITECTURE.md, README.md, CLAUDE.md) and the knowledge base alike — plan-first and evidence-based. A diff is the trigger, but the analysis radius extends to related docs the diff does

• freshness-check

  Validate time-sensitive knowledge — library versions, framework defaults, APIs, best practices — against live registries and the current web before acting on it, because trained knowledge lags reality (a model trained on v1.8 will confidently recommend v1.8 when today's standard is v2.5). USE FOR: c

• go-wild

  Go wild — run an approved plan to completion unattended, building every batch with build and verifying each with verify, in plan order by default and in parallel isolated worktrees only where a batch carries Parallel with, stopping to report when a batch is blocked. USE FOR: "go wild", "run the whol

• grill-me

  Adversarially pressure-tests a decision, plan, belief, or design the user has ALREADY formed — one hard question at a time — until it either hardens or a fatal flaw surfaces. Refuses hand-waving, appeals to authority, and "we'll-figure-it-out-later" answers. USE FOR: "grill me", "stress-test my thin

• impl-strategy

  Create structured implementation plans with a fixed three-section output (What We Are Doing; How We Are Doing It / What Is Out of Scope; Tracking List), written as a plan folder that a later phase executes. It is the architect: it chooses the build strategy and writes it into every batch as ordered

• jira-fetch

  Fetch full Jira ticket details (description, comments, subtasks, linked issues) from a URL or issue key via the read-only REST API, and save them to a markdown file. Credentials are resolved from a lookup chain, never hard-coded. USE FOR: pulling a Jira ticket's full context from a URL or key (e.g.

• lessons-learned

  Standalone lessons-learned log. Captures problems, mistakes, and non-obvious patterns discovered during a session into a `lessons.md` file at a caller-supplied location, with init / read / append modes. USE FOR: initializing, reading, or appending entries to a `lessons.md` log during any working ses

• merge-and-validate

  Safely merges a branch into the current branch and validates the result against the project's Definition of Done, rolling back to the recorded pre-merge state on any failure. USE FOR: merging feature branches, release branches, or hotfixes while ensuring validation gates pass before finishing. DO NO

• objective-advisor

  Turns the agent from a submissive assistant into an objective, high-critical-thinking advisor — targeting cognitive biases, calibrating its own certainty, and pressure-testing ideas without empty contrarianism or polite sycophancy. USE FOR: evaluating a query, proposal, decision, or document where y

• open-knowledge

  The single source of truth for how the repository's knowledge-base/ documentation bundle is organized under the Open Knowledge Format (OKF) v0.2 — frontmatter (including the provenance, trust, and lifecycle families), file and folder naming, folder structure, the reserved index.md and log.md rules,

• review

  Review the change — code health and sound reasoning — before it ships, composing code review, security hardening, simplification, and reasoning checks into one severity-ranked report. USE FOR: the pre-ship review of a branch or local diff, "review this change before merge", the phase between verify

• security-and-hardening

  Hardens code against vulnerabilities. Use when handling user input, authentication, data storage, or external integrations. Use when building any feature that accepts untrusted data, manages user sessions, or interacts with third-party services.

• ship

  Ship it — align documentation content, organize the knowledge base, then merge and validate against the Definition of Done, stopping on any failure. USE FOR: finishing a reviewed branch, "ship it", the final docs-and-merge pass of the definition → build → verify → review → ship path. DO NOT USE FOR:

• thomas

  Hands-on validation skill that refuses to accept any claim that code is "working", "tested", or "done" without personally executing and observing every check. Never assumes, never reasons from probability, and only marks work complete after witnessing passing output first-hand. USE FOR: validating a

• verify

  Verify what was just built — a plan batch or any spec build handed back — by first-hand execution: run its checks, then each gate it names in order, fixing and re-running until every one passes; only witnessed passing output counts. USE FOR: validating a built plan batch before it is marked done, "v

Similar plugins