SDLC Layer 1 Integration Suggestions

SDLC Layer 1 Integration Suggestions

Generated: 2026-02-23
Plan: SDLC Layer Separation Architecture
Layer 1 Scope: Language manifest schema, abstract roles, quality gates, project detection, language standards
Source: Full read of all Layer 1 items from sdlc-layer-candidates-master.md

Plan Reference

The SDLC Layer Separation Architecture defines:

• Layer 0: SDLC-agnostic (harness, SAM pipeline, human touchpoints)
• Layer 1: Language-specific (manifest schema, abstract roles, quality gates, project detection)
• Layer 2: Stack/goal-specific
• ARL: Meta-layer (observe, identify, accumulate, improve)

Layer 1 formalization requires: language manifest schema, abstract role mapping, quality gate declarations, project detection rules, and language standards.

1. language-manifest-template

Path: plugins/development-harness/templates/language-manifest-template.md

Extracted Content

• Skeleton with four sections: Role Fulfillment, Quality Gates, Project Detection, Process Flow Override
• Role Fulfillment: architect, test-designer, code-reviewer, design-spec, linting (agents @plugin:agent, skills /plugin:skill)
• Quality Gates: format, lint, typecheck, test, standards (backtick commands, {files} placeholder)
• Project Detection: markers (config files), source-patterns, test-patterns (glob patterns)
• Process Flow Override: optional, (none) uses default harness flow

Integration Suggestion

Where in plan: Layer 1 formalization — "Language Manifest Schema" section

Amendment: Add explicit reference to the template as the canonical starting point for new language plugins. Require all Layer 1 language plugins to produce a manifest conforming to this template before they can compose with the harness.

Nugget: The template is minimal (28 lines). The full schema with validation rules lives in language-manifest-schema.md. Plan should distinguish: template = quick-start skeleton; schema = validation rules and examples.

2. python3-development

Path: plugins/python3-development/skills/python3-development/SKILL.md

Extracted Content

• ROLE_TYPE identification (orchestrator vs sub-agent) with mandatory echo statement
• Linting Discovery Protocol: git hook tool → CI config → tool detection; format-first workflow
• Type Checker Discovery: basedpyright, pyright, mypy from pre-commit or pyproject.toml
• Quality Gates: format (ruff), lint (ruff), typecheck (detected), test (pytest >80%), modernpython, shebangpython
• Standard Project Structure: packages/, tests/, pyproject.toml, hatchling
• Agent orchestration: python-cli-architect, python-pytest-architect, python-code-reviewer, python-cli-design-spec, swarm-task-planner
• Pre-Delegation Checklist: read orchestration guide, identify workflow, plan agent chain, define scope, set success criteria
• Anti-pattern: DO NOT pre-gather data for agents (CoVe bypass)

Integration Suggestion

Where in plan: Layer 1 — "Reference Implementation: Python" subsection

Amendment: Add python3-development as the canonical Layer 1 reference. Plan should require other language plugins to provide equivalent: (1) Role Fulfillment mapping to harness abstract roles, (2) Quality Gates with discovery protocol, (3) Project Detection markers (pyproject.toml, setup.py, setup.cfg), (4) Orchestration guide for agent chaining.

Nugget: The Linting Discovery Protocol (git hook → CI → fallback) is a reusable pattern. Extract as a Layer 1 standard: "Language plugins MUST implement a discovery sequence before executing quality gates."

3. python-development-orchestration

Path: plugins/python3-development/skills/python3-development/references/python-development-orchestration.md

Extracted Content

• Five workflow patterns: TDD, Feature Addition, Code Review, Refactoring, Debugging
• Agent selection: python-cli-architect (default), stdlib-scripting (last resort), python-pytest-architect, python-code-reviewer, python-cli-design-spec, swarm-task-planner
• Decision tree for scripts: internet access? → python-cli-architect; no uv? → stdlib-scripting
• Quality Gates: holistic-linting, pytest, modernpython, shebangpython
• Anti-patterns: don't write Python as orchestrator; don't skip validation; don't mix agent contexts

Integration Suggestion

Where in plan: Layer 1 — "Workflow Patterns" (new subsection)

Amendment: Add workflow pattern taxonomy as a Layer 1 concept. Each language plugin should document: (1) TDD-equivalent, (2) Feature Addition, (3) Code Review, (4) Refactoring, (5) Debugging — with agent chains and quality gates per pattern.

Nugget: The "Context to pass: file paths, outcomes, user requirements only. Do not pass file contents" rule is critical for CoVe. Include in Layer 1 standards: "Orchestrators pass paths and outcomes; agents discover and verify."

4. perl-development

Path: plugins/perl-development/ (SKILL.md, agents: perl-script-developer, perl-script-auditor, perl-cli-architect)

Extracted Content

• Perl 5.30+ standards: strict, warnings, autodie; feature (say, state, signatures)
• Agents: perl-script-developer, perl-script-auditor, perl-cli-architect
• Skills: perl-development, perl-cpan-ecosystem, perl-environment-setup, perl-testing, perl-lint, perl-validate
• Patterns: Path::Tiny, Try::Tiny, Getopt::Long, POD; security (taint, IPC::Open3 for safe system calls)
• No explicit language manifest file; agents and skills exist but no harness-compatible manifest

Integration Suggestion

Where in plan: Layer 1 — "Language Plugin Inventory" or "Manifest Gap Analysis"

Amendment: Add task: "Create references/language-manifest.md for perl-development plugin using language-manifest-template." Map: architect → perl-cli-architect; test-designer → perl-testing skill or new agent; code-reviewer → perl-script-auditor; design-spec → perl-cli-architect; linting → perl-lint.

Nugget: Perl project detection markers: Makefile.PL, Build.PL, cpanfile, META.json. Quality gates: perl -c (syntax), perlcritic, prove (tests).

5. bash-development

Path: plugins/bash-development/ (SKILL.md, agents: bash-script-developer, bash-script-auditor)

Extracted Content

• Bash 5.1+ standards: set -euo pipefail; trap-based error handling
• Agents: bash-script-developer, bash-script-auditor
• Skills: bash-51-features, bash-52-features, bash-53-features, bash-development, bash-lint, bash-logging, bash-portability, bash-testing
• Patterns: pure-bash alternatives, argument parsing template, variable best practices
• No explicit language manifest; version-specific features (5.1, 5.2, 5.3) as separate skills

Integration Suggestion

Where in plan: Layer 1 — "Language Plugin Inventory" or "Manifest Gap Analysis"

Amendment: Add task: "Create references/language-manifest.md for bash-development plugin." Map: architect → bash-script-developer (or new bash-architect); test-designer → bash-testing; code-reviewer → bash-script-auditor; design-spec → bash-script-developer; linting → bash-lint. Project detection: no standard markers (scripts often standalone); consider: .bashrc, Makefile with bash shebang, or directory with *.sh.

Nugget: Bash quality gates differ: format (shfmt if present), lint (shellcheck), typecheck (N/A), test (bats or custom). Plan should allow "typecheck: (none)" for languages without static typing.

6. agent-creator

Path: .claude/skills/agent-creator/SKILL.md

Extracted Content

• Phase 1–7 workflow: Discovery, Requirements, Template Selection, Adaptation, Generation, Validation, Integration
• Role archetypes: Researcher, Planner/Architect, Coder, Creator, Tester, Reviewer, DevOps, Auditor, Context Gatherer, Optimizer, Domain Expert
• Standard vs Role-Based Contract: user-facing (flexible) vs orchestrated (DONE/BLOCKED)
• Template selection via AskUserQuestion; adapt from existing agents or archetypes

Integration Suggestion

Where in plan: Layer 1 — "Abstract Role → Agent Mapping" (cross-reference)

Amendment: Add cross-reference: "When creating language-specific agents (e.g., python-cli-architect), use agent-creator skill. Role archetypes (Coder, Reviewer, Architect) align with harness abstract roles (architect, code-reviewer, design-spec)."

Nugget: The Domain Expert archetype is the pattern for language-pro agents (javascript-pro, typescript-pro, c-systems-programmer). Plan should state: "Language-pro agents are Domain Expert archetype instances with language-specific skills and quality checklists."

7. agent-schema

Path: .claude/skills/agent-creator/references/agent-schema.md

Extracted Content

• YAML frontmatter spec: name, description, model, tools, skills, permissionMode, hooks, mcpServers, memory, etc.
• Required: name (lowercase, hyphens, max 64), description (action verbs, triggers, keywords, max 1024)
• YAML multiline bug: do NOT use >-, | for descriptions (indexer breaks)
• Validation rules before saving

Integration Suggestion

Where in plan: Layer 1 — "Agent Definition Standards" (new subsection)

Amendment: Clarify distinction: (1) Language manifest (development-harness) maps abstract roles to agents; (2) Agent schema (agent-creator) defines agent frontmatter. Both are Layer 1 concerns. Plan should require language plugin agents to conform to agent-schema (name, description, skills) and be referenced in the language manifest.

Nugget: Agent schema does not define "language manifest for agents" — that's the development-harness language manifest. The agent-schema is for individual agent .md files. Avoid conflating the two in plan wording.

8. agent-templates

Path: .claude/skills/agent-creator/references/agent-templates.md

Extracted Content

• Role-Based Contract Archetypes: Researcher, Planner/Architect, Coder, Creator, Tester, Reviewer, DevOps, Auditor, Context Gatherer, Optimizer, Domain Expert
• All use skills: subagent-contract for DONE/BLOCKED signaling
• Domain Expert: model: inherit, color, skills: {{domain_skill}}
• Supervisor co-prompt templates per archetype

Integration Suggestion

Where in plan: Layer 1 — "Abstract Roles and Archetype Mapping"

Amendment: Add mapping table: harness abstract role → agent-creator archetype. Example: architect → Planner/Architect; test-designer → Tester; code-reviewer → Reviewer; design-spec → Planner/Architect or Coder; linting → (skill, not agent). Plan should state that language manifests resolve harness roles to agents that implement these archetypes.

Nugget: The Coder archetype with stack-specific skills (e.g., skills: subagent-contract, python3-development) is the pattern for language-specific implementation agents. Plan should document this as the standard for language plugin "coder" role fulfillment.

9. agent-examples

Path: .claude/skills/agent-creator/references/agent-examples.md

Extracted Content

• Real implementations: plugin-assessor, code-review, contextual-ai-documentation-optimizer, doc-drift-auditor, plugin-refactor:refactor-skill, context-gathering
• Role-based examples: Coder (Next.js+Supabase), Coder (Python TUI), Formatter with hooks
• Patterns: description with triggers, permissionMode, color, phased workflow, self-verification

Integration Suggestion

Where in plan: Layer 1 — "Implementation Reference" (nugget only)

Amendment: No direct plan amendment. Use as reference when creating language-specific agents.

Nugget: The Coder (Python TUI) example shows skills: subagent-contract, python3-development — the exact pattern for language manifest code-reviewer/coder fulfillment. Plan should cite this as the reference implementation for stack-specific Coder agents.

10. skill-research-process

Path: .claude/skills/skill-research-process/SKILL.md

Extracted Content

• 3-stage process: Initialize → Research → Integrate
• Quality Gate 1: Category verification (distinct, complete)
• Quality Gate 2: Anti-hallucination checkpoint (every claim cited, no training data)
• Quality Gate 3: Final validation (links, structure)
• Citation format required; MCP tool selection (Ref > exa > WebFetch)

Integration Suggestion

Where in plan: Layer 1 — "Quality Gates for Meta-Artifacts" (skill/agent creation)

Amendment: Add that skill-research-process quality gates (especially Gate 2: anti-hallucination) apply when creating language-specific skills. Language standards documented in skills must have cited sources. Plan should reference this for "language standards" documentation quality.

Nugget: Gate 2 pattern — "Every factual claim has a cited source" — should extend to language manifest quality gate commands. When a manifest declares lint: \ruff check {files}``, the plan could require that the command be verified against project config (discovery protocol) rather than assumed.

11. optimize-claude-md

Path: .claude/skills/optimize-claude-md/SKILL.md

Extracted Content

• 8 optimization principles: positive framing, motivation, concrete examples, front-loaded priorities, concise language, explicit format control, strategic XML tagging, structural enforcement
• RT-ICA pre-check, CoVe post-check in delegation template
• Phases: Validate → Measure → Delegate → Verify → Measure → Report → Apply
• Applies to CLAUDE.md, SKILL.md, agent definitions

Integration Suggestion

Where in plan: Layer 1 — "Quality Standards for AI-Facing Content" (ARL overlap)

Amendment: Add that language plugin documentation (CLAUDE.md, SKILL.md, agent prompts) should meet optimize-claude-md principles. When formalizing Layer 1, run optimize-claude-md on language manifest schema docs and template to ensure clarity for AI consumption.

Nugget: The 8 principles apply to language manifest documentation. Plan should require: language manifest schema and template are optimized for AI parsing (front-loaded structure, explicit format, concrete examples per language).

12. javascript-pro

Path: .claude/agents/javascript-pro.md

Extracted Content

• ES2023+, Node.js 20+ specialist
• Quality checklist: ESLint, Prettier, tests >85%, JSDoc, bundle size, async error handling, no var
• Project detection: read package.json, build config, module system
• Operational protocol: read config → analyze patterns → implement → verify

Integration Suggestion

Where in plan: Layer 1 — "Language-Pro Agent Pattern" (new subsection)

Amendment: Add javascript-pro as a Layer 1 language-pro agent. It implements the Domain Expert archetype with: (1) quality checklist (ESLint, Prettier, tests), (2) project detection (package.json), (3) language standards (ES2023+). Plan should require a language manifest entry mapping code-reviewer or architect to @javascript-pro when a TypeScript/JavaScript plugin is created.

Nugget: javascript-pro has no explicit language manifest. If a "javascript-development" or "typescript-development" plugin is created, it should include a manifest with: markers: package.json; quality gates: eslint, prettier, test; role fulfillment including javascript-pro or typescript-pro.

13. typescript-pro

Path: .claude/agents/typescript-pro.md

Extracted Content

• TypeScript 5.0+ specialist; strict mode always
• Quality checklist: strict mode, no any, 100% type coverage, type-only imports, source maps, declaration files
• Project detection: tsconfig.json, package.json, build configs
• Initialization protocol: read config → assess type patterns → identify framework → check lint config

Integration Suggestion

Where in plan: Layer 1 — "Language-Pro Agent Pattern" (same as javascript-pro)

Amendment: typescript-pro extends the pattern with type-specific gates (type coverage, strict mode). Plan should note: TypeScript/JavaScript plugins can share project detection (package.json) but differ on typecheck (tsc for TS, N/A for JS).

Nugget: typescript-pro has hardcoded path in memory: /home/ubuntulinuxqa2/repos/claude_skills/.claude/agent-memory/typescript-pro/. Plan should include backlog item: "Replace hardcoded agent memory paths with {AGENT_MEMORY_DIR} or relative paths" (consistent with development-harness service-docs-maintainer fix).

14. c-systems-programmer

Path: .claude/agents/c-systems-programmer.md

Extracted Content

• C99/C11, POSIX, Valgrind, GDB specialist
• Quality focus: memory management, pointer safety, error handling, thread safety
• Tool recommendations: GCC/Clang -Wall -Wextra, Valgrind, GDB, clang-tidy
• No explicit project detection; no language manifest

Integration Suggestion

Where in plan: Layer 1 — "Language-Pro Agent Pattern"

Amendment: c-systems-programmer is a Domain Expert for C. Plan should add: "C development plugin (if created) would use markers: Makefile, CMakeLists.txt, Makefile.am; quality gates: format (clang-format), lint (clang-tidy), typecheck (N/A or -Wall), test (ctest/make test)."

Nugget: C projects often lack a single canonical config file. Project detection for C may need: Makefile, CMakeLists.txt, configure.ac, or *.c file presence. Plan should document that some languages have ambiguous detection and allow fallback to "source pattern match only."

Summary: Plan Amendments to Add

Cross-References for Plan

• language-manifest-schema: plugins/development-harness/skills/development-harness/references/language-manifest-schema.md
• role-resolution-protocol: plugins/development-harness/skills/development-harness/references/role-resolution-protocol.md
• development-harness CLAUDE.md: plugins/development-harness/CLAUDE.md (composition model, Voltron diagram)