CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

evaluate

Use when the user wants proof that finished work meets its acceptance criteria, not just that tests pass: 'verify it is actually done', 'check it against the spec', 'evaluate the run'. Runs mechanical checks, a semantic review per criterion, and multi-model consensus when uncertain. With execution.auto_evolve on (the default), a rejected result with a Seed continues into a Ralph fix loop; pass auto_evolve: false to verify only.

Quality

88%

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

SKILL.md
Quality
Evals
Security

Quality

Content

88%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 highly actionable, well-sequenced protocol with complete tool-call blocks, an executable fallback command, and per-outcome feedback loops. The only real weakness is readability of Step 4, where the sentinel-contract validation rules compress into one dense paragraph that would benefit from being split or extracted into a reference file.

Suggestions

Break Step 4 into short bullet points (valid sentinel criteria, observer-child delegation, no-child fallback, chained-Ralph handling) instead of one run-on paragraph.

Extract the RFC sentinel/observer-contract validation protocol into a bundled reference file (e.g., references/observer-contract.md) and keep a two-line summary in the body.

Normalize the step numbering (replace 2.5 with a renumbered step) so the sequence reads cleanly.

DimensionReasoningScore

Conciseness

The body is dense and imperative, explains no general concepts Claude already knows, and every section carries protocol-specific detail — efficient overall. It is not a 5 because Step 4 is a single ~200-word run-on paragraph mixing sentinel validation, child-session delegation, and polling rules that could be tightened or broken up; not a 3 because there is no genuinely unnecessary explanation, just compressed density.

4 / 5

Actionability

Fully executable guidance: a complete `ouroboros_start_evaluate` argument block ready to copy, a copy-paste-ready `uvx ... python -c ...` fallback command for the evaluator prompt, concrete probe classes (misleading_output, hung_command, stale_state, dirty_worktree), and an explicit per-outcome next-action map. This matches the top anchor for copy-paste-ready commands covering the common cases.

5 / 5

Workflow Clarity

Clear sequenced steps (Session, Artifact, Acting verification, Start, Observe, Present) with explicit validation checkpoints — Stage 1 authority gates approval, the sentinel contract must pass canonical v1 validation fail-closed — and explicit feedback loops for every rejection branch (fix failures then re-evaluate, ooo run, ooo evolve, ooo interview). It is not a 4 because no validation gaps are apparent; the sequence and checkpoints are complete, including a defined outcome for each failure mode.

5 / 5

Progressive Disclosure

The skill has no bundle files and is well-organized into clear sections (Stage authority, Instructions, Steps, Fallback, Footer), with external references ($OUROBOROS_AGENTS_DIR/evaluator.md, src/ouroboros/agents/evaluator.md, .ouroboros/mechanical.toml) one level deep and clearly signaled in their Fallback section. It is not a 5 because the Step 4 observer-contract protocol is the kind of detailed procedural content that would ideally live in a bundled reference file rather than inlined as a wall of prose — a minor organization gap.

4 / 5

Total

18

/

20

Passed

Description

83%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 strong description: explicit 'what' and 'when', with naturally-quoted trigger phrases users would actually say. The main deduction is that capability detail is partly deferred to internal jargon (Seed, Ralph, auto_evolve) and a few natural trigger synonyms are absent.

Suggestions

Enumerate one or two representative mechanical checks (e.g., 'lint, build, tests') in the description itself instead of deferring the term entirely to the body.

Add one or two more natural trigger phrasings users would say, such as 'review the result' or 'does it meet the requirements'.

DimensionReasoningScore

Specificity

Names several concrete actions — 'Runs mechanical checks, a semantic review per criterion, and multi-model consensus when uncertain' and 'a rejected result with a Seed continues into a Ralph fix loop' — which fits the anchor for several specific actions. It is not a 5 because 'mechanical checks' are not itemized (lint/build/test only appear in the body) and the tail leans on internal jargon; not a 3 because more than 1-2 concrete actions are explicitly stated.

4 / 5

Completeness

Explicitly answers both: what ('Runs mechanical checks, a semantic review per criterion, and multi-model consensus') and when ('Use when the user wants proof that finished work meets its acceptance criteria, not just that tests pass') with concrete quoted trigger phrases — a direct match for the top anchor.

5 / 5

Trigger Term Quality

Includes three naturally-quoted user phrasings — 'verify it is actually done', 'check it against the spec', 'evaluate the run' — plus 'acceptance criteria', giving good keyword coverage. It falls short of comprehensive (5) because common variations like 'review', 'is this correct', or 'does it meet requirements' are missing, and internal jargon (Ralph, auto_evolve) adds trigger noise.

4 / 5

Distinctiveness Conflict Risk

The niche — verifying finished work against acceptance criteria rather than just tests passing — is clear, with distinct trigger phrases. Minor overlap risk remains because 'verify it is actually done' could also match a generic code-review or verification request, so it is mostly distinct rather than minimal conflict risk.

4 / 5

Total

17

/

20

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.

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

15

/

16

Passed

Repository

Q00/ouroboros

Path

skills/evaluate/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.