Skip to content
Back to skills

2490 Grooming 2026 02 23 381502e2

ASecurity

**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) ---

  • 9 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 11, 2026
researchpythonnodegitapidevopsdocumentation

Works with

  • terminal
  • cli
  • api
  • mcp

Security analysis

A100/100

Scanned October 11, 2026

npx -y skills add tools-only/X-Skills --skill 2490-grooming-2026-02-23_381502e2 --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of 2490 Grooming 2026 02 23 381502e2?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for 2490 Grooming 2026 02 23 381502e2
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/tools-only-2490-grooming-2026-02-23-381502e2/badge)](https://www.skillsdirectory.com/skills/tools-only-2490-grooming-2026-02-23-381502e2)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

SKILL.md
# 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

| Item | Priority | Fact-Check | RT-ICA | Research Found | Skills | Agents | Blockers |
|------|----------|------------|--------|----------------|--------|--------|----------|
| conventional-commits: Fix CHANGELOG references | P2 | V:1 | APPROVED | CHANGELOG, plugin structure | plugin-creator | — | None |
| dasel: Reconcile 265 -f flag occurrences | P2 | V:1 | APPROVED | dasel docs, plugin.json | dasel-reference | data-explorer | Manifest drift |
| verification-gate: Remove 95% confidence claim | P2 | R:1 | APPROVED | SKILL.md, research-foundations | verification-gate | — | None |
| development-harness: Remove hardcoded path | P2 | V:1 | APPROVED | service-docs-maintainer.md | development-harness | — | None |
| llamafile: Fix HuggingFace model URLs | P2 | Pre-checked | APPROVED | Backlog citations | — | — | Verify mozilla-ai repos |
| prompt-optimization: Fix unreachable refs | P2 | V:1 | APPROVED | SKILL.md, references | prompt-optimization | — | None |
| brainstorming-skill: Remove orphaned refs | P2 | V:1 | APPROVED | SKILL.md, references | brainstorming-skill | — | None |
| uv: Fix incorrect script paths in README | P2 | I:1 | APPROVED | README, python3-development | uv | — | Path audit needed |
| plugin-creator: Remove dead code | P2 | V:1 | APPROVED | plugin_validator, normalize_frontmatter | plugin-creator | — | BLE001 in normalize_frontmatter |
| github_project_setup: Projects V2 status | P2 | V:1 | APPROVED | projects-v2.md, GraphQL | gh | — | None |
| Validate carbonyl Terminal Browser | Ideas | — | BLOCKED | — | agent-browser | — | TTY + network |
| plugin-creator: Plugin Validation MCP | Ideas | — | APPROVED | plugin_validator, scripts | plugin-creator | plugin-assessor | None |
| hallucination-detector: Hallucination Audit MCP | Ideas | — | APPROVED | hallucination-audit-stop.js | hallucination-detector | — | JS→Python or wrap |
| gitlab-skill: GitLab CI & GLFM MCP | Ideas | — | APPROVED | validate_glfm, sync_gitlab_docs | gitlab-skill | — | None |
| holistic-linting: Linting Orchestration MCP | Ideas | — | APPROVED | holistic_lint.py | holistic-linting | — | None |
| development-harness: Task Status MCP | Ideas | — | APPROVED | implementation_manager | development-harness | — | None |
| summarizer: File Metrics MCP | Ideas | — | APPROVED | file_metrics.py | summarizer | — | None |
| the-rewrite-room: Document Analysis MCP | Ideas | — | APPROVED | file_metrics.py --json | the-rewrite-room | — | Overlap with summarizer |
| fastmcp-creator: MCP Meta-Tooling MCP | Ideas | — | APPROVED | connections, evaluation | fastmcp-creator | — | None |
| python3-development: Python Toolchain MCP | Ideas | — | APPROVED | mkdocs sync, uv sync | python3-development | — | None |
| dasel: Structured Data Query MCP | Ideas | — | APPROVED | dasel CLI | dasel | data-explorer | Subprocess vs reimplement |

---

## 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](https://github.com/bitflight-devops/stateless-agent-methodology)

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.