CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

orbit

Running autonomous loops for nexus-autoloop. Generates script sets from goals, designs operation contracts, audits live loops, and recovers state — runners that complete reliably.

Quality

76%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./.claude/skills/orbit/SKILL.md

The canonical home for this skill is orbit in simota/agent-skills

SKILL.md
Quality
Evals
Security

Quality

Content

73%Weight 40%Scale 1-5

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

A dense, highly actionable orchestration skill with an exemplary phased workflow table and strong one-level-deep reference navigation. Its main weaknesses are redundancy — the HTML comment block, restated circuit-breaker/token-bound detail, and duplicate Output Contract headings — which inflates token cost without adding guidance.

Suggestions

Delete or gut the HTML comment block (CAPABILITIES_SUMMARY / COLLABORATION_PATTERNS / BIDIRECTIONAL_PARTNERS): it nearly verbatim duplicates the Collaboration, Trigger Guidance, and Handoff Tokens sections while spending ~40 lines of tokens.

Keep only the rule plus a pointer where inline detail restates reference files — e.g., the Circuit Breaker section and Token bound paragraph already say "Detail -> reference/core-defaults.md / reference/resilience-patterns.md", so the repeated mechanics can live solely in those files.

Merge the two sections titled "Output Contract" ("### Output Contract" under Output and Handoffs and the later "## Output Contract" tier/style section) into one location to remove duplicated structure and conflicting navigation.

DimensionReasoningScore

Conciseness

The body is telegraphic and table-driven with no explanations of concepts Claude already knows, but carries real redundancy: the ~40-line HTML comment block (CAPABILITIES_SUMMARY / COLLABORATION_PATTERNS / BIDIRECTIONAL_PARTNERS) duplicates the Collaboration and Trigger Guidance sections; Circuit Breaker and token-bound detail are restated inline after pointing to reference/core-defaults.md and reference/resilience-patterns.md; and two sections are both titled "Output Contract". Mostly efficient but includes unnecessary duplication — not a 4, since the redundancy exceeds minor trims.

3 / 5

Actionability

Exact thresholds (DEDUP_WINDOW=5, NEXUS_LOOP_SUMMARY <= 180 chars, floor 20,000 tokens), concrete commands (bash -n on every *.sh, recover.sh --reset-circuit), exact field lists, token spellings, and deterministic dispatch rules make the guidance mostly executable. It is not a 5 because the fully copy-paste-ready content (templates, verify.sh wiring) lives in referenced files rather than the body, leaving minor gaps.

4 / 5

Workflow Clarity

The INTAKE -> CONTRACT -> ... -> LEARN pipeline is given as an ordered table with a required action and a key rule per phase, plus explicit validation checkpoints and feedback loops: VERIFY blocks HANDOFF on any failure, PRE_FLIGHT aborts on [PREFLIGHT:FAIL], DONE requires an evidence gate with rollback note, and RF-01 is non-skippable. This matches the anchor for clear sequence with explicit validation and error-recovery loops.

5 / 5

Progressive Disclosure

References are one level deep and well signaled via the per-recipe "Read First" column, per-phase reference lists, and a full Reference Map with "Read this when" guidance. It is not a 5 because no bundle files are present to verify, and detail explicitly delegated to reference/core-defaults.md, reference/resilience-patterns.md, and reference/failure-catalog.md is still restated inline, leaving the ~230-line body heavier than a clean overview.

4 / 5

Total

16

/

20

Passed

Description

66%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 terse, third-person description with a clear named-tool niche and several concrete capabilities, held back by the absence of any explicit "Use when..." trigger clause. Adding natural trigger phrases (loop, runner, autoloop, stuck loop, loop audit) would raise both completeness and trigger-term quality.

Suggestions

Add an explicit trigger clause, e.g., "Use when the user mentions nexus-autoloop, autonomous loops, loop runners, stuck or oscillating loops, or asks to audit/recover a live loop."

Include natural synonyms and artifact names users would actually say (runner, loop goal, goal.md, runner.log, loop audit) rather than only ecosystem jargon like "operation contracts".

Clarify what "script sets" concretely are (runner + bootstrap/recover/verify/notify scripts) so the primary capability is unambiguous.

DimensionReasoningScore

Specificity

"Generates script sets from goals, designs operation contracts, audits live loops, and recovers state" lists several specific concrete actions with only minor coverage gaps (e.g., what the script sets contain is left implicit). It is not a 5 because coverage is not comprehensive and leans partly on domain jargon ("operation contracts").

4 / 5

Completeness

The "what" is clear (generate, design, audit, recover), but there is no "Use when..." clause or equivalent explicit trigger guidance — the "when" is only weakly implied by "for nexus-autoloop". Per rubric guidance, a missing explicit trigger clause caps completeness at 3; it is not a 2 because the "what" is concrete and a weak "when" is present.

3 / 5

Trigger Term Quality

Terms like "autonomous loops", "nexus-autoloop", "audits live loops", and "recovers state" give good keyword coverage for the niche, but natural user variations such as "runner", "stuck loop", "loop recovery", or artifact names (goal.md, runner.log) are missing. Sits above anchor 3 but short of the synonym-level coverage of anchor 5.

4 / 5

Distinctiveness Conflict Risk

"Running autonomous loops for nexus-autoloop" names a specific tool and niche, giving mostly distinct triggers with minimal conflict risk against generic document/data skills. It is not a 5 because "generates script sets... audits" could still brush against generic automation or CI skills in a large ecosystem.

4 / 5

Total

15

/

20

Passed

Validation

100%

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

simota/agent-skills

Path

.claude/skills/orbit/SKILL.md

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.