claude-code

Delegate coding to Claude Code CLI (features, PRs).

Quality

50%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Fix and improve this skill with Tessl

tessl review fix ./skills/autonomous-ai-agents/claude-code/SKILL.md

Quality

Content

50%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The content is exceptionally actionable with concrete, executable commands and clear orchestration patterns, but it is also very verbose and monolithic, repeating Claude Code reference material inline rather than splitting it into one-level-deep reference files. The main improvement is trimming known reference detail and moving exhaustive flag/command tables into separate referenced files.

Suggestions

Trim exhaustive reference tables of CLI flags, slash commands, and keyboard shortcuts that Claude already knows; keep only the orchestration-specific guidance and the few flags essential to Hermes integration.

Move the complete CLI flags reference and slash-command catalog into a separate file (e.g., references/cli-reference.md) linked from a concise overview, achieving true one-level-deep progressive disclosure.

For the --dangerously-skip-permissions and other destructive flows, replace timing-based sleeps with explicit validation checks (e.g., verify the expected prompt text via capture-pane before sending the next keystroke) to add real feedback loops.

Dimension	Reasoning	Score
Conciseness	The body is a ~730-line monolith enumerating every CLI flag, slash command, and keyboard shortcut in full tables, much of which Claude already knows about Claude Code; this padded reference material matches the level-1 anchor of verbosity that explains concepts and capabilities Claude already has.	1 / 3
Actionability	Guidance is highly concrete and copy-paste ready: executable terminal() invocations with real flags (--max-turns, --output-format json, --allowedTools), specific tmux sequences, and exact dialog-handling keystrokes, matching the level-3 anchor of fully executable commands.	3 / 3
Workflow Clarity	Multi-step processes (PTY launch, dialog handling, parallel instances) are clearly sequenced with sleeps and capture-pane monitoring checkpoints, but feedback loops for risky/destructive operations like --dangerously-skip-permissions are timing-based rather than explicit validate-fix-retry checks, capping this at level 2.	2 / 3
Progressive Disclosure	There are no bundle/reference files, so everything lives inline in one well-sectioned document; the flag and slash-command reference tables are content that could be split into separate files but is not, fitting the level-2 anchor of some structure with content that should be separate kept inline.	2 / 3
	Total	8 / 12 Passed

Description

50%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is concise and identifies the target tool plus a couple of use cases, but it lacks explicit 'Use when...' trigger guidance and concrete action verbs. Adding natural trigger terms and a clear activation clause would raise the weaker dimensions.

Suggestions

Add an explicit 'Use when...' clause stating when Claude should invoke this skill (e.g., 'Use when you need to run Claude Code non-interactively, automate coding tasks, or review PRs').

Replace the abstract 'Delegate coding' with concrete actions (e.g., 'Run one-shot coding tasks in print mode, drive multi-turn sessions via tmux, and automate PR review').

Include natural user-facing keywords like 'coding agent', 'code review', 'refactoring', and 'CI automation' to improve trigger coverage.

Dimension	Reasoning	Score
Specificity	Names the tool ('Claude Code CLI') and a couple domains ('features, PRs') but the action 'Delegate coding' is abstract rather than a list of concrete actions, matching the level-2 anchor that names a domain and some actions but not comprehensively.	2 / 3
Completeness	It states what the skill does ('Delegate coding to Claude Code CLI') but provides no explicit 'Use when...' trigger, and the judging guidelines cap completeness at 2 when explicit trigger guidance is missing.	2 / 3
Trigger Term Quality	'Claude Code CLI' and 'PRs' are relevant keywords, but common natural variations users would actually say ('coding agent', 'code review', 'refactoring', 'automate coding') are missing, so coverage is partial rather than complete.	2 / 3
Distinctiveness Conflict Risk	Tying it to 'Claude Code CLI' gives a recognizable niche, but 'Delegate coding' is generic enough that it could overlap with other coding-agent skills, so it is not yet a fully distinct trigger set.	2 / 3
	Total	8 / 12 Passed

Validation

75%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 12 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (746 lines); consider splitting into references/ and linking	Warning
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	12 / 16 Passed

Repository: NousResearch/hermes-agent
Path: skills/autonomous-ai-agents/claude-code/SKILL.md
Commit: 5be99b6

Reviewed: about 15 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.