start-work

Execute a Prometheus work plan in Codex with Boulder state, evidence ledger updates, worktree discipline, parallel subagents, and Stop-hook continuation. Use after planning when the user says start work, execute plan, continue plan, resume plan, or asks to run a .omo/plans plan.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

77%

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

An actionable, well-sequenced orchestration skill with strong validation checkpoints, but it is dense and padded with duplicated guidance and lacks any progressive disclosure to offload detail into reference files.

Suggestions

De-duplicate the subagent reliability guidance (back-off timing, WORKING/BLOCKED signals, 'treat a running child as alive', fallback conditions) — state it once and reference it, instead of repeating it near-verbatim across the tool-compatibility and subagent-reliability sections.

Move the OpenCode→Codex tool-translation table, the 9 ultraqa class definitions, and the boulder.json/DoneClaim/AdversarialVerify JSON schemas into separate reference files under ./references/ and link to them one level deep to reduce the monolithic body.

Quarantine the time-sensitive surface-version note ('verified 2026-07-11: only fork_turns/message/task_name') into a clearly marked compatibility/deprecated section so stale date markers do not dilute the core instructions.

Dimension	Reasoning	Score
Conciseness	Mostly efficient but padded: the subagent back-off/WORKING/BLOCKED/fallback guidance is duplicated nearly verbatim across 'Codex Harness Tool Compatibility' and 'Codex Subagent Reliability', 'this section wins' is repeated three times, and an inline time-sensitive date ('verified 2026-07-11') is not quarantined in a deprecated section.	2 / 3
Actionability	Highly concrete and copy-paste ready — exact commands ('git worktree add', 'git worktree list --porcelain', the web-terminal-visual-qa invocation), explicit JSON schemas (boulder.json, DoneClaim/AdversarialVerify), and required ledger fields — meeting the fully executable anchor.	3 / 3
Workflow Clarity	A clear Phase 1–5 + Completion sequence with explicit validation gates (the five-gate Phase 4, Global Review and Debugging Gate) and feedback loops (non-confirmed verdict → re-dispatch; review failure → debugging skill → rerun), appropriate for the destructive/PR/merge operations involved.	3 / 3
Progressive Disclosure	Well-sectioned with headers but monolithic: no bundle files exist and content that could be split out (the OpenCode→Codex tool table, the 9 ultraqa classes, the JSON schemas, the DoneClaim contract) is all inline in a ~200-line body, rather than one-level-deep references.	2 / 3
	Total	10 / 12 Passed

Description

100%

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

A precise, third-person description that states concrete capabilities and an explicit 'Use when...' trigger clause covering natural user phrasings. It distinguishes itself clearly from neighboring skills.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete components — 'Boulder state, evidence ledger updates, worktree discipline, parallel subagents, and Stop-hook continuation' — matching the 'lists multiple specific concrete actions' anchor rather than the single-domain score 2.	3 / 3
Completeness	Clearly answers what (execute a Prometheus work plan with the listed components) and when ('Use after planning when the user says...'), with an explicit trigger clause so it is not capped at 2.	3 / 3
Trigger Term Quality	Explicit natural triggers users would say — 'start work, execute plan, continue plan, resume plan, or asks to run a .omo/plans plan' — giving good coverage of common variations rather than jargon-only.	3 / 3
Distinctiveness Conflict Risk	A clearly bounded niche (Prometheus/Codex work-plan execution with .omo/plans) with distinct triggers, making accidental firing for unrelated skills unlikely.	3 / 3
	Total	12 / 12 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: code-yeongyu/lazycodex
Path: plugins/omo/skills/start-work/SKILL.md
Commit: 895b70c

Reviewed: 4 days 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.