2490 Grooming 2026 02 23 381502e2

Backlog Grooming Report

Date: 2026-02-23 Items groomed: 21 (ungroomed items from 2026-02-21 baseline) Arguments: all Scope: P2 items not in prior report + Ideas MCP server items (2026-02-23)

Summary

Individual Manifests

conventional-commits: Fix CHANGELOG references to nonexistent files (P2)

RT-ICA: APPROVED — file existence check is straightforward.

Fact-check: VERIFIED — CHANGELOG references docs/examples.md and [1.0.0]: https://github.com/owner/conventional-commits/releases/tag/v1.0.0. Plugin has no docs/ folder; release URL is placeholder. Dead references confirmed.

Supporting Skills: /plugin-creator (validation patterns) Prior Work: plugins/conventional-commits/ — CHANGELOG.md lines 27-31 Blockers: None Suggested First Steps: 1) Remove or fix CHANGELOG references; 2) Audit skill cross-references in plugin; 3) Run plugin_validator

dasel: Reconcile 265 -f flag occurrences with reference documentation (P2)

RT-ICA: APPROVED — documentation and manifest fix are DERIVABLE.

Fact-check: VERIFIED — dasel plugin.json has no hooks key; 265+ -f occurrences across skill files. Reference docs need audit against official dasel CLI.

Supporting Skills: /dasel-reference, /data-exploration Related Agents: @data-explorer Prior Work: plugins/dasel/ — format-patterns.md, format-recipes.md Blockers: Run auto_sync_manifests.py --reconcile for hook registration Suggested First Steps: 1) Reconcile manifest; 2) Fetch dasel CLI --help; 3) Audit -f usage vs official docs

verification-gate: Remove unsubstantiated 95% confidence claim (P2)

RT-ICA: APPROVED — claim removal is trivial; citation addition would require methodology.

Fact-check: REFUTED as unsubstantiated — "5% cost for 95% reliability improvement" appears in SKILL.md line 397 and research-foundations.md line 234. No supporting data, citation, or methodology. Per CLAUDE.md verification protocol, remove or rephrase as qualitative.

Supporting Skills: /verification-gate Prior Work: plugins/verification-gate/skills/verification-gate/SKILL.md, references/research-foundations.md Blockers: None Suggested First Steps: 1) Replace "5% cost for 95% reliability" with qualitative phrasing (e.g., "low overhead for high reliability"); 2) Fix code fence language specifiers; 3) Update stale dates

development-harness: Remove hardcoded machine path and fix version drift (P2)

RT-ICA: APPROVED — path and version fixes are AVAILABLE from description.

Fact-check: VERIFIED — Hardcoded path /home/ubuntulinuxqa2/repos/claude_skills/.claude/agent-memory/service-docs-maintainer/ in plugins/development-harness/agents/service-docs-maintainer.md line 164.

Supporting Skills: /development-harness Prior Work: plugins/development-harness/ Blockers: None Suggested First Steps: 1) Replace with {AGENT_MEMORY_DIR} or relative path; 2) Audit version refs; 3) Reconcile role table with actual roles

llamafile: Fix HuggingFace model URLs (P2)

RT-ICA: APPROVED — backlog already fact-checked 2026-02-21.

Fact-check: Pre-checked — org Mozilla → mozilla-ai; model repo names need verification against actual mozilla-ai org.

Prior Work: Backlog citations; plugins/llamafile/skills/llamafile/SKILL.md Blockers: Verify current mozilla-ai model repo names (may have changed since 2026-02-21) Suggested First Steps: 1) WebFetch HuggingFace mozilla-ai org; 2) Update URLs; 3) Remove fabricated repo names

prompt-optimization: Fix unreachable reference files and raw JSX/MDX markup (P2)

RT-ICA: APPROVED — structural fixes.

Fact-check: VERIFIED — SKILL.md does not link to accessing_online_resources.md or whats-new-claude-4.5.md; context-windows.md exists. Raw JSX/MDX needs conversion.

Supporting Skills: /prompt-optimization-claude-45 Prior Work: plugins/prompt-optimization-claude-45/skills/prompt-optimization-claude-45/ Blockers: None Suggested First Steps: 1) Add links from SKILL.md to unreachable refs or remove; 2) Convert JSX/MDX to standard markdown

brainstorming-skill: Remove orphaned bibliography entry and cross-reference headings (P2)

RT-ICA: APPROVED — structural cleanup.

Fact-check: VERIFIED — SKILL.md references multiple refs; backlog claims orphaned bibliography and cross-reference headings. Grep shows references exist; need to identify which are orphaned.

Supporting Skills: /brainstorming-skill Prior Work: plugins/brainstorming-skill/skills/brainstorming-skill/ Blockers: None Suggested First Steps: 1) Trace all ref links; 2) Remove orphaned entries; 3) Fix headings pointing to removed content

uv: Fix incorrect script paths in README (P2)

RT-ICA: APPROVED — path audit needed.

Fact-check: INCONCLUSIVE — README describes "compatibility wrapper" and references python3-development. Specific incorrect paths not located in first 80 lines; full README audit required.

Supporting Skills: /uv, /python3-development Prior Work: plugins/uv/README.md (256 lines) Blockers: Identify exact incorrect paths Suggested First Steps: 1) Compare README script paths to actual file locations; 2) Consider README size reduction (thin wrapper)

plugin-creator: Remove dead code and triplicated regex (P2)

RT-ICA: APPROVED — code quality fixes.

Fact-check: VERIFIED — normalize_frontmatter.py line 140 has # noqa: BLE001; fix_tool_formats.py has skipped_files (used); plugin_validator has multiple skipped/sum usages (contextual). Triplicated regex and dead skipped list need grep audit.

Supporting Skills: /plugin-creator, /refactor-skill Related Agents: @refactor-executor Prior Work: plugins/plugin-creator/scripts/, planning/plugin-validator-qa-report.md Blockers: BLE001 in normalize_frontmatter — address per CLAUDE.md linting policy Suggested First Steps: 1) Grep for triplicated regex patterns; 2) Remove dead code; 3) Fix BLE001 with specific exception; 4) Audit HK005 treatment

github_project_setup.py: add GitHub Projects V2 status field updates (P2)

RT-ICA: APPROVED — API and reference docs AVAILABLE.

Fact-check: VERIFIED — projects-v2.md documents updateProjectV2ItemFieldValue mutation (line 137); projectV2Fields query for field ID discovery. PyGithub or GraphQL required.

Supporting Skills: /gh Prior Work: .claude/skills/gh/scripts/github_project_setup.py, references/projects-v2.md Blockers: None Suggested First Steps: 1) Add project_app Typer sub-app; 2) Implement project update-status --project-number N --issue-number N --status "In Progress"; 3) Use GraphQL for field value update

Validate carbonyl Terminal Browser (Ideas)

RT-ICA: BLOCKED — TTY and network requirements not satisfiable in current environment.

Prior Work: agent-browser skill Blockers: Needs TTY (Inappropriate ioctl for device); DNS blocked in restricted env Suggested First Steps: Run on host with TTY and unrestricted network

plugin-creator: Plugin Validation & Scaffolding MCP (Ideas)

RT-ICA: APPROVED — scripts exist, wrapping is DERIVABLE.

Supporting Skills: /plugin-creator Related Agents: @plugin-assessor Prior Work: plugin_validator.py, auto_sync_manifests.py, create_plugin.py, normalize_frontmatter.py Tools to expose: validate_plugin, create_plugin, sync_manifests, normalize_frontmatter, list_validation_errors Suggested First Steps: 1) Create plugins/plugin-creator/mcp/server.py; 2) Wrap CLI entry points as FastMCP tools

hallucination-detector: Hallucination Audit MCP (Ideas)

RT-ICA: APPROVED — core logic exists in JS.

Prior Work: hallucination-audit-stop.js Blockers: Reimplement in Python or wrap JS via subprocess Tools: audit_text, audit_file Suggested First Steps: 1) Decide Python reimplement vs Node subprocess; 2) Create mcp/server.py or server.js

gitlab-skill: GitLab CI & GLFM Validation MCP (Ideas)

RT-ICA: APPROVED — scripts exist.

Prior Work: validate_glfm.py, sync_gitlab_docs.py Tools: validate_glfm, sync_docs, list_glfm_errors Suggested First Steps: 1) Create mcp/server.py; 2) Wrap scripts as tools

holistic-linting: Linting Orchestration MCP (Ideas)

RT-ICA: APPROVED — holistic_lint.py exists.

Prior Work: holistic_lint.py Tools: lint_files, list_lint_errors, auto_fix Suggested First Steps: 1) Create mcp/server.py; 2) Expose structured lint results

development-harness: Task Status MCP (Ideas)

RT-ICA: APPROVED — implementation_manager CLI exists.

Prior Work: implementation_manager.py (python3-development plugin) Tools: query_tasks, list_pending, update_task_status, parse_task_file Suggested First Steps: 1) Create mcp/server.py under development-harness; 2) Wrap implementation_manager

summarizer: File Metrics & Summarization MCP (Ideas)

RT-ICA: APPROVED — file_metrics.py exists.

Prior Work: the-rewrite-room file_metrics.py (count, scan with --json) Tools: count_tokens, scan_files, summarize_file Note: Overlap with the-rewrite-room Document Analysis MCP Suggested First Steps: 1) Locate summarizer plugin file_metrics; 2) Create mcp/server.py

the-rewrite-room: Document Analysis MCP (Ideas)

RT-ICA: APPROVED — file_metrics.py has --json support.

Prior Work: file_metrics.py (count, scan --json) Tools: count_document_metrics, scan_documents, compare_versions Suggested First Steps: 1) Create mcp/server.py; 2) Coordinate with summarizer if consolidating

fastmcp-creator: MCP Meta-Tooling MCP (Ideas)

RT-ICA: APPROVED — connections.py, evaluation.py, get_environment.py exist.

Prior Work: FastMCP plugin scripts Tools: test_mcp_connection, run_evaluation, get_mcp_environment, validate_mcp_config Suggested First Steps: 1) Create mcp/server.py; 2) Wrap existing utilities

python3-development: Python Development Toolchain MCP (Ideas)

RT-ICA: APPROVED — mkdocs sync, uv release sync scripts exist.

Prior Work: python3-development plugin scripts Tools: sync_mkdocs, sync_uv_releases, check_python_environment Suggested First Steps: 1) Create mcp/server.py; 2) Wrap scripts

dasel: Structured Data Query MCP (Ideas)

RT-ICA: APPROVED — dasel CLI available.

Research first: Subprocess vs reimplement in Python Prior Work: dasel plugin, data-explorer agent Tools: query_data, transform_data, convert_format, install_dasel Suggested First Steps: 1) Evaluate subprocess wrapping vs Python reimplementation; 2) Create mcp/server.py

Fact-Check Results

Refuted Claims

• verification-gate: "5% cost for 95% reliability improvement" — REFUTED as unsubstantiated. No methodology, data, or citation. Remove or rephrase as qualitative.

Inconclusive Claims

• uv: "README contains incorrect script paths" — INCONCLUSIVE. README is 256 lines; specific incorrect paths not yet identified. Full path audit needed.

Verified Claims

• conventional-commits: CHANGELOG references docs/examples.md and placeholder release URL — VERIFIED dead (docs/ folder absent)
• dasel: 265+ -f occurrences, plugin.json missing hooks — VERIFIED
• development-harness: Hardcoded path in service-docs-maintainer.md — VERIFIED
• llamafile: Pre-fact-checked 2026-02-21 (HuggingFace org/repos)
• prompt-optimization: Unreachable reference files — VERIFIED
• plugin-creator: noqa BLE001, dead code patterns — VERIFIED
• github_project_setup: updateProjectV2ItemFieldValue in projects-v2.md — VERIFIED

RT-ICA Results

BLOCKED Items

• Validate carbonyl Terminal Browser: TTY and network requirements not satisfiable in restricted environment.

APPROVED Items

• 20 items — conditions verified or DERIVABLE from codebase.

Cross-Item Findings

Shared Dependencies

• MCP Ideas: All 10 Ideas MCP items depend on FastMCP or similar MCP server pattern. agentskill-kaizen is reference implementation (kaizen-analysis server.py).
• plugin-creator: Referenced by conventional-commits (validation), plugin-creator dead code, Plugin Validation MCP.
• dasel: P2 -f reconciliation and Ideas Structured Data Query MCP both touch dasel plugin — consider batching.

Suggested Groupings

• Plugin cleanup wave: conventional-commits, verification-gate, development-harness, prompt-optimization, brainstorming-skill, uv, plugin-creator dead code — all structural/doc fixes.
• MCP server wave: Batch MCP Ideas by plugin (plugin-creator, hallucination-detector, gitlab-skill, etc.) — shared FastMCP patterns.
• dasel: Do P2 -f reconciliation and Ideas MCP in same session — same plugin.

Research Gaps

• uv README paths: Need systematic comparison of README script paths vs actual file locations.
• dasel -f flag: Official dasel CLI documentation for format/selector flags.
• MCP subprocess vs reimplement: dasel, hallucination-detector need decision on wrapping binary/JS vs Python reimplementation.

Methodology: stateless-agent-methodology