sdlc

Plugin sdlc from marketplace ai-sdlc in bibutikoley/ai-sdlc-claude. Commands read from the repo's manifest — 1 month ago.

Skills (20)

• babysit

  Stage 5 (Deploy) PR shepherding for the AI-native SDLC. Sweep a PR's unresolved review comments and failing checks, address them, push fixes, and repeat until the PR is green and waits only on code owner approval. Use on /sdlc:babysit or when asked to get a PR to green / address review comments.

• eval-add

  Stage 4 (Test) continuous evals. Turn a production incident, a fixed bug, or a real recent task into a permanent eval case under evals/cases/ so the agent configuration (CLAUDE.md, skills, hooks, REVIEW.md) is regression-tested in CI. Use on /sdlc:eval-add or when asked to add an eval / regression c

• incident

  Stage 6 (Maintain) of the AI-native SDLC. Diagnose an anomaly, alert, ticket, or channel message with read-only tools, write the finding as an incident intent.md that re-enters the loop, and update lessons.md. Use on /sdlc:incident, when a control band is breached, when asked to investigate an alert

• intent

  Stage 1 (Plan) of the AI-native SDLC. Turn an idea, ticket, or problem statement into a committed intent.md through an analyst-style brainstorm. Use when someone has an idea, describes a pain point, wants to "write this up", or says /sdlc:intent. Works for non-engineers.

• metrics

  Compute and explain the AI-native SDLC's leading and lagging indicators from Git history, PR metadata and CI (intent time-to-commit, intent→spec, spec→plan, plan→merge, rework counts, first-pass CI success, review time, gate waits, breach→intent). Use on /sdlc:metrics or when leadership asks how the

• plan

  Stage 3 (Build) entry of the AI-native SDLC. In plan mode, interview the engineer against an accepted spec.md and produce a committed plan.md (files that change, order of work, risks, options not taken, proof) before any code is written. Use on /sdlc:plan or when asked to plan a change that has a sp

• brand-guidelines

  The organisation's brand and content standards applied to specs and UI work. Use when writing spec.md for anything customer- or employee-facing, when producing UI copy, emails, notifications, or visual components, and when reviewing front-end changes.

• compliance-policy

  Regulatory and change-management obligations as constraints on specs, plans, releases and reviews. Use when writing spec.md or plan.md, when a change affects regulated processes, audit trails, records retention, customer communications, or when preparing a release, and during the Compliance review p

• secure-api-review

  Apply the API security standard. Use whenever creating or modifying an external-facing endpoint, reviewing API code, or generating an OpenAPI spec.

• security-policy

  The organisation's security policy as constraints on specs, plans, code and reviews. Use when writing spec.md or plan.md, when a change touches authentication, authorisation, secrets, PII or regulated data, network egress, dependencies, or infrastructure, and during the Security review pass.

• ux-standards

  Interaction, accessibility and content-structure standards for anything users touch. Use when writing spec.md for a user-facing change, designing flows or states, building or reviewing UI, or comparing an implementation to a mock.

• release

  Stage 5 (Deploy) release of the AI-native SDLC. Prepare a release (changelog from merged changes, checks, staging rehearsal), deploy through tiered environments via MCP or the configured commands, and stop at the production gate until a release manager authorises. Use on /sdlc:release or when asked

• review

  Stage 5 (Deploy) AI review of the AI-native SDLC. Apply REVIEW.md's passes (bugs, security, compliance against spec.md and plan.md) to a diff, branch or PR, rank findings by severity, cap nits, and emit a machine-readable tally. Use on /sdlc:review, when asked to review a PR/diff/branch, or non-inte

• rollback

  Roll back an environment to the previous release using the single rehearsed rollback command, then record the incident intent. Use on /sdlc:rollback, when a deploy fails, when a post-deploy control band is breached, or when asked to revert a release.

• sdlc-loop

  Shared reference for the AI-native SDLC loop used by every /sdlc command. Load when working with intent.md, spec.md, plan.md, sdlc/changes, sdlc/incidents, REVIEW.md, release gates, or sdlc.config.json, or when asked how the SDLC loop, its artifacts, gates or roles work.

• setup

  Install the AI-native SDLC loop into a product repository. Asks about CI platform (GitHub/GitLab/both), tracker, chat, monitoring, build/test/lint commands and protected paths, then scaffolds sdlc/, sdlc.config.json, REVIEW.md, CI workflows, evals, monitoring, MCP connectors and the CLAUDE.md sectio

• spec

  Stage 2 (Design) of the AI-native SDLC. Read an accepted intent.md and produce a requirements-and-design spec.md constrained by the organisation's policy skills (security, compliance, brand, UX), with flagged concerns first. Use when an intent is accepted, when asked for a spec/requirements/design f

• status

  Show where every change and incident is in the AI-native SDLC loop (which artifact exists, its status, who owns the next gate), plus open PRs linked to changes. Use on /sdlc:status or when asked "where is change X", "what is waiting on me", or for a loop overview.

• tracker-sync

  Keep the legacy tracker (Jira, YouTrack, GitLab issues, GitHub issues) and the Markdown artifacts linked according to sdlc.config.json sourceOfTruth (repo | tracker | linked). Read the record at session start, write outcomes and commit SHAs back through the tracker's MCP connector. Load whenever an

• verify

  Stage 4 (Test) of the AI-native SDLC. Give Claude a way to check its own work: run the configured build/test/lint with quantifiable targets, drive the failing-test-first bug-fix flow (with the test-file lock), and the screenshot loop for UI. Use on /sdlc:verify, when asked to verify/prove a change w

Subagents (4)

• adversarial-reviewer

  Independent confidence gate between autonomous stages of the SDLC loop. Given an artifact an agent just produced (spec.md, plan.md, an incident diagnosis, a proposed PR), tries to break it and returns CONTINUE or ESCALATE with reasons. Used by CI jobs that chain stages headlessly.

• code-simplifier

  Strips needless complexity from a finished diff without changing behaviour. Use after the main agent completes a change and before opening the PR. Runs the tests before and after.

• researcher

  Explores the codebase, docs, and history to answer a focused question and reports back concisely, so the main session's context is not flooded with file dumps. Read-only.

• verifier

  Runs the app and the configured checks and verifies the change works before the session reports done. Fresh context, so the verdict is not coloured by the assumptions that produced the code. Report only; never fixes.

Hooks (9)

• SessionStart

  "${CLAUDE_PLUGIN_ROOT}"/hooks/scripts/session-context.sh

• PreToolUse

  "${CLAUDE_PLUGIN_ROOT}"/hooks/scripts/protected-paths.sh

• PreToolUse

  "${CLAUDE_PLUGIN_ROOT}"/hooks/scripts/test-guard.sh

• PreToolUse

  "${CLAUDE_PLUGIN_ROOT}"/hooks/scripts/secrets-guard.sh

• PreToolUse

  "${CLAUDE_PLUGIN_ROOT}"/hooks/scripts/production-gate.sh

• PreToolUse

  "${CLAUDE_PLUGIN_ROOT}"/hooks/scripts/secrets-guard.sh

• PreToolUse

  "${CLAUDE_PLUGIN_ROOT}"/hooks/scripts/role-gate.sh

• PostToolUse

  "${CLAUDE_PLUGIN_ROOT}"/hooks/scripts/format-on-edit.sh

• Stop

  "${CLAUDE_PLUGIN_ROOT}"/hooks/scripts/plan-sync.sh

Similar plugins