Tessl • The management layer for AI agent skills

A lean, expert-calibrated checklist with repo-specific tooling references and genuine validation discipline (reproduce the failing seed, rerun the CI-equivalent shard before declaring victory). The only meaningful gap is the absence of executable examples for the two most load-bearing commands — the CI shard invocation and `capture_debug_errors_during` usage.

Suggestions

Add a one-line copy-paste example of the CI-equivalent shard/full test command (or a pointer to where its exact syntax is found in the job log), which would move actionability toward fully executable.

Add a minimal usage snippet for `capture_debug_errors_during` (capture scope + asserting exact messages), since it is the least self-evident helper named in the gates.

Dimension	Reasoning	Score
Conciseness	The body is 29 lines of dense imperative rules with zero padding or explanation of concepts Claude already knows — every bullet ("Reproduce reported flakes with the failing `--rng-seed`", "Restore global state with `restore_on_out_of_scope`") earns its tokens. Even the longest CI bullet carries only specific, load-bearing detail, matching the 'lean and efficient; every token earns its place' anchor.	5 / 5
Actionability	Guidance is highly concrete and repo-specific — named paths (`data/mods/TEST_DATA`), flags (`--rng-seed`), and helpers (`restore_on_out_of_scope`, `clear_map()`, `capture_debug_errors_during`) — but no copy-paste commands or code snippets are given (e.g., the shard-command invocation syntax or a minimal `capture_debug_errors_during` usage example). This fits 'mostly executable guidance with minor gaps' rather than the fully copy-paste-ready 5 anchor.	4 / 5
Workflow Clarity	The gates embed explicit ordering and validation checkpoints — "before accepting the fix", "Run that same shard command on the final rebased HEAD before declaring the fix validated", "Before pushing test fixes... run both the relevant focused test and the CI-equivalent shard/full command" — which is strong verify-before-proceed discipline. It falls short of the 5 anchor because there is no explicit error-recovery loop for what to do when the CI-equivalent rerun still fails, and the bullets are a checklist rather than a fully sequenced workflow.	4 / 5
Progressive Disclosure	The skill is under 50 lines, has no external references to deeper material, and is organized into two clear, well-labeled sections (Gates, Cleanup) plus a trigger line. Per the scoring notes, a simple skill under 50 lines with no need for external references scores 5 with just well-organized sections; the referenced paths (`data/mods/TEST_DATA`) are repo paths in the target project, not bundle files, and no bundle directories exist.	5 / 5
	Total	18 / 20 Passed

Description

70%Weight 40%Scale 1-5

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 specific, niche-scoped description with concrete actions and natural trigger terms, undermined only by the absence of an explicit 'Use when...' clause, which caps completeness at 3. Adding the body's existing trigger sentence to the description would resolve the main weakness.

Suggestions

Append an explicit trigger clause, e.g. 'Use when adding, changing, or reviewing tests in the Cataclysm-BN codebase, or when CI test jobs fail.' — this would lift completeness from 3 to 4-5 and add natural trigger terms like 'flaky' and 'CI test failures'.

Surface the CI-failure reproduction capability (reproducing flakes with the failing `--rng-seed`, rerunning the CI shard command) in the description to close the specificity coverage gap.

Dimension	Reasoning	Score
Specificity	The description lists several concrete actions — "Write or review Cataclysm-BN tests", "avoid flakes", "use deterministic fixtures", and "clean up global state" — which is well above the 'names domain, minimal actions' anchor. It stops short of the 5 anchor because prominent capabilities from the body (e.g., reproducing CI shard failures with a failing `--rng-seed`) are not surfaced, leaving minor coverage gaps.	4 / 5
Completeness	The 'what' is clear (write/review CBN tests with flake avoidance, deterministic fixtures, and global-state cleanup), but the description contains no 'Use when...' clause or equivalent trigger guidance — the 'when' is only weakly implied by "Write or review". Per the judging guidelines, a missing explicit 'when' caps completeness at 3, even though the body's first line ('Use when adding or changing automated tests') supplies it.	3 / 5
Trigger Term Quality	Natural terms are present: "tests", "flakes", "fixtures" — a user debugging a flaky test or writing CBN tests would plausibly say these. A few common variations are missing ("flaky test", "unit tests", "CI test failures"), so it fits the 'good keyword coverage; a few natural terms missing' anchor rather than the comprehensive 5.	4 / 5
Distinctiveness Conflict Risk	"Cataclysm-BN tests" carves out a clear niche with distinct triggers (repo-specific fixtures, `--rng-seed`, shard commands); it would not fire for generic test-writing or other skills. This matches the 'clear niche with distinct triggers; minimal conflict risk' anchor.	5 / 5
	Total	16 / 20 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: cataclysmbn/Cataclysm-BN
Path: .agents/skills/test-quality/SKILL.md
Revision: 9d4abdd

Reviewed: 1 month ago

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.

test-quality

Table of Contents

Is this your skill?