CtrlK
BlogDocsLog inGet started
Tessl Logo

codex-first

Claude Code-only work routing: delegate implementation, fixing, exploratory subagents, rebasing, and PR merging/landing to Codex CLI while Claude specifies, decides, reviews, and verifies. Direct hands-on delegation requires ANTHROPIC_BASE_URL to be unset or neither loopback nor gorillaclaw.sheep-coho.ts.net. Codex-backed autoreview is always allowed and preferred, independent of environment.

61

Quality

72%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Critical

Do not install without reviewing

Fix and improve this skill with Tessl

tessl review fix ./skills/codex-first/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

77%

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

The body is highly actionable and workflow-clear, with executable Codex commands, explicit gates, and verification checklists, but it is dense and monolithic — prose rationale and long-run details could be trimmed or offloaded to keep the core overview lean.

Suggestions

Trim or move the economics rationale and minification-incident lore into a short 'Rationale' reference file so the main flow stays lean.

Split the Liveness watchdog and Prompt Contract sections into one-level-deep reference files (e.g., WATCHDOG.md, PROMPT-CONTRACT.md) signaled from the overview to improve progressive disclosure.

Tighten prose-heavy bullets (e.g., the harness-visibility paragraph) into imperative directives without losing the operational signal.

DimensionReasoningScore

Conciseness

Mostly high-signal operational guidance, but includes explanatory padding — the economics rationale ('Claude tokens metered + expensive; Codex flat-rate') and the minification-incident anecdote — that could be tightened; not the lean level-3 ideal.

2 / 3

Actionability

Fully executable, copy-paste-ready commands with pinned flags and model ('command codex exec --yolo -C <repo> -m gpt-5.6-sol ...'), plus the resume, watchdog, and launcher install scripts — concrete rather than pseudocode.

3 / 3

Workflow Clarity

Clear sequences with explicit validation gates ('rebase→resolve→push→CI attach+green→land'), feedback loops ('after 2 failed rounds, take over'), and a coordinator verification checklist for landed PRs.

3 / 3

Progressive Disclosure

Well-organized into labeled sections, but it is a monolithic 200+ line single file with no reference offloading; the watchdog and prompt-contract details could be split into one-level-deep reference files.

2 / 3

Total

10

/

12

Passed

Description

67%

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 specific and distinctive, naming concrete delegation categories and the complementary Claude-side role, but lacks an explicit natural-language 'Use when' trigger clause and leans on infrastructure jargon that weakens trigger-term quality.

Suggestions

Add an explicit 'Use when...' clause with natural user phrasings (e.g., 'Use when you need to rebase, land/merge a PR, implement a frozen spec, or fix CI/lint failures') to lift completeness to 3.

Move environment/router internals (ANTHROPIC_BASE_URL host checks, gorillaclaw.sheep-coho.ts.net) out of the description into the body's Hard gate so trigger terms stay natural.

Expand trigger coverage with common variations users actually say — 'refactor', 'write tests', 'migrate', 'bump dependencies' — instead of jargon like 'exploratory subagents'.

DimensionReasoningScore

Specificity

Lists multiple concrete actions — 'delegate implementation, fixing, exploratory subagents, rebasing, and PR merging/landing to Codex CLI while Claude specifies, decides, reviews, and verifies' — matching the multi-action anchor rather than the single-domain level 2.

3 / 3

Completeness

The 'what' is explicit and a conditional gate gives a partial 'when', but there is no explicit 'Use when…' trigger clause, which the rubric caps at 2.

2 / 3

Trigger Term Quality

Natural task terms (rebase, land/merge PR, implement, fix) appear but are diluted by infrastructure jargon ('ANTHROPIC_BASE_URL', 'gorillaclaw.sheep-coho.ts.net', 'exploratory subagents', 'autoreview') and missing common variations like 'refactor' or 'write tests'.

2 / 3

Distinctiveness Conflict Risk

A clear niche — routing hands-on coding work to Codex CLI with Codex-backed autoreview — with distinct triggers unlikely to fire for unrelated skills.

3 / 3

Total

10

/

12

Passed

Validation

93%

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

Validation15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

referenced_paths_exist

Referenced path issues: 3 missing

Warning

Total

15

/

16

Passed

Repository
steipete/agent-scripts
Reviewed

Table of Contents

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.