sdlc-plugins

Software Development Lifecycle (SDLC) plugins for AI coding agents, packaged using the Agent Plugins spec (v1.0.0).

What's Included

This plugin bundles a set of Agent Skills—reusable, discoverable instructions that an agent loads on demand based on the task at hand.

Skill	Description
sdlc-behavior-principles	Behavioral guardrails for planning, implementing, and reviewing changes: surface assumptions, keep solutions simple, make surgical diffs, and require verifiable checks.
verify-task	Enforces explicit proof of correctness (requirement, test, regression, diff, unproven) before a task is declared done.
code-documenter	Generates and maintains OKF-compliant knowledge bases documenting a codebase's modules, classes, functions, architecture, and configuration.
legacy-code-discovery	Orchestrates evidence-graded discovery of a legacy estate before a migration: charter, coverage ledger, phases, gates, and an OKF decision dossier.
repo-reconnaissance	Read-only structural survey: repository topology, stacks and versions, build/test/run entry points, deployable units, exclusions, and a scalable shard plan.
system-cartography	Maps components, typed dependency edges, entry points, end-to-end flows, integration contracts, and data ownership without assuming a tech stack.
migration-seams	Converts the map into migration decisions: risk scoring, coupling hotspots, behavioral parity hazards, seams, readiness, and slice sequencing.

The discovery skills are stack-neutral and scale from a single repository to monorepos and multi-repo estates.

No MCP servers are configured in mcp.json at this time.

Installation

Add this repository as a plugin source in a client that supports the Agent Plugins spec, then enable sdlc-plugins. Consult your client's documentation for how to register a plugin by Git URL or local path.

GitHub Copilot CLI

This repository is also published as a plugin marketplace for GitHub Copilot CLI:

copilot plugin marketplace add dhanoopbhaskar/sdlc-plugins
copilot plugin install sdlc-plugins@sdlc-plugins

Usage

Once installed, skills are picked up automatically by the agent when their description matches the current task—no manual invocation is required. For example:

Ask the agent to plan, implement, or review a change and it will apply the sdlc-behavior-principles guardrails.
Ask the agent to confirm a task is complete and it will apply the verify-task checklist.
Ask the agent to document a codebase and it will apply the code-documenter workflow.
Ask the agent to map a legacy system before a migration and it will apply legacy-code-discovery, scaling its survey with repo-reconnaissance, system-cartography, and migration-seams.

Project Structure

plugin.json               # Plugin manifest (required)
mcp.json                  # MCP servers configuration
CHANGELOG.md              # Project changelog
skills/                   # Agent Skills, one directory per skill
  <skill-name>/SKILL.md
  <skill-name>/references/ # Supporting material loaded on demand
scripts/
  validate_skills.py      # Validates skill frontmatter, links, and version metadata
tests/
  test_validate_skills.py # Tests the validator and its frontmatter parser
  fixtures/okf-bundle/    # Minimal OKF bundle used by the conformance test
.github/plugin/
  marketplace.json        # Plugin marketplace manifest for GitHub Copilot CLI
.github/workflows/
  validate.yml            # CI validation on push and pull request

Validation

Run the repository checks locally:

python3 scripts/validate_skills.py
python3 -m unittest discover -s tests -v

The script validates Skill frontmatter, naming, relative and image links (including angle-bracket destinations, reference definitions, heading anchors, and bundle-absolute links inside OKF bundles, while ignoring fenced/indented code and HTML comments), manifest shape and version metadata, and generated-actor versions. The tests cover the validator, its frontmatter parser, its link, manifest, and version checks, and a minimal OKF bundle fixture. The same checks run in CI on every push and pull request.

When bumping the plugin version, update any concrete generated.by example actors in the skills; the validator fails when they drift from plugin.json.

Contributing

See CHANGELOG.md for release history. Follow Keep a Changelog and Semantic Versioning conventions when proposing changes.

License

MIT