327 Backlog 193d131d

Backlog

Tracked features, ideas, and deferred work. Each item lives in .claude/backlog/.

P0 - Must Have

• Reduce session-start context load via rules path-scoping and disable-model-invocation #198

P1 - Should Have

• gitlab-skill: Remove hardcoded corporate URL

• bash-development: Fix bash-53-features inaccuracies and task_output bug

• clang-format: Fix broken YAML frontmatter

• agent-orchestration: Remove phantom /is-it-done command references

• perl-development: Fix shell injection vulnerability in example template

• hallucination-detector: Fix dead backtick evidence marker and multi-occurrence "because" bug

• python3-development: Fix 3 malformed frontmatter files

• the-rewrite-room: Fix nonexistent script reference and 6 broken links

• fastmcp-creator: Add citations for 1200+ lines of FastMCP 3.x API documentation

• Consolidate validate_frontmatter.py into plugin_validator.py

• Validate and verify orchestrator-discipline plugin hooks and processes

• SAM: Error Recovery / Rollback Procedures #199

• SAM: Human Escalation Criteria #200

• SAM: Timeout/Stall Detection #201

• SAM: Artifact Schema Validation #202

• SAM: Scope Creep Detection #203

• Meta-Process Capture — Expert Panel Dataset Builder #204

• SAM Extension — Integrate ARL General Theory #205

• ARL Skill Development #206

• Extract claude-plugin-lint to standalone PyPI package #207

• P1: plugin-validator pre-commit output is too noisy #208

• P1: plugin-validator UX and coverage gaps #209

• daily-releases: Task AI analysis step to model=haiku subagent and fix duplicate draft releases #196

• Fix pre-existing CI check failures on main

• Backlog system redesign: GitHub Issues as source of truth with local cache

• backlog.py: unify issue body template and add missing structured fields

• Enhance skill-research-process for CLI tool skills #197

• Create ty skill for python3-development plugin #210

P2 - Could Have

• conventional-commits: Fix CHANGELOG references to nonexistent files #211

• dasel: Reconcile 265 -f flag occurrences with reference documentation #212

• verification-gate: Remove unsubstantiated 95% confidence claim #213

• development-harness: Remove hardcoded machine path and fix version drift #214

• llamafile: Fix HuggingFace model URLs (wrong org name + fabricated repos) #215

• prompt-optimization: Fix unreachable reference files and raw JSX/MDX markup #216

• brainstorming-skill: Remove orphaned bibliography entry and cross-reference headings #217

• uv: Fix incorrect script paths in README #218

• plugin-creator: Remove dead code and triplicated regex #219

• Add PR003/PR004 test coverage to plugin registration validator #220

• kaizen: MCP consolidation analysis #221

• SAM: Parser regex false positive on "## Task Summary Statistics" #222

• SAM: Replace validate-task-file.sh with Python validator #223

• SAM: Parallel Execution Details #224

• SAM: Multi-Model Strategy #225

• SAM: Audit Trail / Observability #226

• SAM: Partial Success Handling #227

• SAM: Context Size Management #228

• SAM: Conflicting Review Findings #229

• Multi-session build state lost during context compaction #230

• /plugin-dev:create-plugin workflow lacks intra-phase parallelism tracking #231

• Background agent result deduplication after compaction #232

• /plugin-dev:create-plugin Phase 6 validation should batch fixes by file, not by finding #233

• Plan artifact diverges from implementation without update mechanism #234

• Evaluate scikit-learn dependency weight for agentskill-kaizen cluster_sessions tool #235

• github_project_setup.py: add milestone close command

• work-backlog-item: accept #N in close and resolve routing

• github_project_setup.py: add GitHub Projects V2 status field updates #236

• Configurable Token Thresholds for plugin_validator.py #237

• SAM: Cost/Token Management #238

• SAM: Team Coordination Protocols #239

• SAM: External System Integration Patterns #240

• SAM: Migration Strategy Guide #241

• SAM: Training/Onboarding Materials #242

• SAM: Non-Code Workflow Guidance #243

• Carbonyl Browser Integration for Claude Code #195

• Validate is-fast for Web Content Extraction #244

• Validate agent-browser for Web Automation #245

• Validate carbonyl Terminal Browser #246

• work-backlog-item: Implement --language and --stack routing #247

• refresh-research: Implement --layer filtering logic #248

• ARL human-probing: Implement skill/agent #194

• evaluate-sdlc-layers: Run and fix findings #249

• SDLC layers: Cross-reference audit #250

• TEST: Backlog script verification

• TEST: Resolve verification

• conventional-commits: Fix SKILL.md frontmatter, broken links, and structure #251

• Fix pre-existing linting errors in gitlab_context.py #252

Ideas

• plugin-creator: Plugin Validation & Scaffolding MCP #253
• hallucination-detector: Hallucination Audit MCP #254
• gitlab-skill: GitLab CI & GLFM Validation MCP #255
• holistic-linting: Linting Orchestration MCP #256
• development-harness: Task Status MCP #257
• summarizer: File Metrics & Summarization MCP #258
• the-rewrite-room: Document Analysis MCP #259
• fastmcp-creator: MCP Meta-Tooling MCP #260
• python3-development: Python Development Toolchain MCP #261
• dasel: Structured Data Query MCP #262
• Item title #263
• Research Resources #264

Completed

• Resolve 48 pre-existing ty (Astral type checker) diagnostics
• Resolve plugin-validator pre-existing errors to make CI gate blocking
• Resolve manifest drift to make CI sync check blocking
• Replace requests with httpx in all scripts
• Enhance swarm-task-planner with multi-source synthesis
• Add context compliance checking
• SAM: Artifact Versioning Strategy
• Create ecosystem-researcher agent

Format Guide

See .claude/docs/backlog-item-groomed-schema.md for groomed item structure.