CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

clarity-gate

Pre-ingestion verification for epistemic quality in RAG systems. Ensures documents are properly qualified before entering knowledge bases. Produces CGD (Clarity-Gated Documents) and validates SOT (Source of Truth) files.

Quality

60%

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 ./plugins/AI-Agents-Safe-Coding-Skills/skills/clarity-gate/SKILL.md

The canonical home for this skill is clarity-gate in administrakt0r/AI-Agents-Safe-Coding-Skills

SKILL.md
Quality
Evals
Security

Quality

Content

46%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.

The body is information-dense and genuinely actionable (commands, templates, fix patterns) but is buried in a monolithic 680-line SKILL.md padded with changelogs, marketing tables, and duplicated validation tables. The biggest structural problem is progressive disclosure: every detail is inlined and the referenced spec/scripts don't exist in the bundle, so the skill neither stays lean nor successfully delegates.

Suggestions

Move the changelog, 'What's New in v2.1' table, E-TB validation tables, hash/claim-ID algorithm details, and SOT specifics into reference files (e.g., references/validation-codes.md, references/changelog.md), keeping SKILL.md as a lean overview with clearly signaled one-level-deep links.

Either bundle the referenced scripts/claim_id.py, scripts/document_hash.py, and docs/CLARITY_GATE_FORMAT_SPEC.md, or inline the minimal content actually needed to run them — currently the copy-paste commands point at files that don't exist in the skill bundle.

Add an explicit numbered end-to-end workflow (e.g., 1. scan against the 9 points, 2. apply fixes/markers, 3. compute document hash, 4. generate CGD, 5. run Round A, 6. run Round B, 7. emit final report) with validation checkpoints, and remove the duplicated E-TB01-07 table that appears in two sections.

DimensionReasoningScore

Conciseness

The ~680-line body carries substantial padding: a 60+ line changelog, a 'What's New in v2.1' feature table, the E-TB01-07 validation table duplicated verbatim in both 'Validation Codes' and 'SOT Validation', and a 'Related Projects' promo table. Not 1 because most content (codes, formats, script algorithms) is skill-specific knowledge Claude doesn't have, not re-explanation of known concepts.

2 / 5

Actionability

Concrete, executable guidance throughout: exact commands with expected outputs ('python scripts/claim_id.py --test', 'document_hash.py --verify my-doc.cgd.md'), a complete CGD YAML template, specific fix patterns ('PROJECTED:', '[assuming X]'), and a fill-in report format. Not 5 because the invoked scripts/claim_id.py, scripts/document_hash.py, and docs/CLARITY_GATE_FORMAT_SPEC.md are not present in the bundle, so some copy-paste guidance dead-ends.

4 / 5

Workflow Clarity

The verification hierarchy diagram, 9 points, and HITL round templates imply a flow, but no explicit end-to-end ordered procedure exists (read doc → run 9 points → compute hash → emit CGD → Rounds A/B → report). Checkpoints appear in fragments (HITL confirmations, --verify, W-HC01 partial-field detection) yet the sequence connecting them is implicit, leaving the reader to assemble the order.

3 / 5

Progressive Disclosure

The bundle contains no references/, scripts/, or assets/ directories, yet the body invokes scripts/*.py, cites docs/CLARITY_GATE_FORMAT_SPEC.md, and inlines ~680 lines of spec detail (validation codes, hash algorithms, claim-ID collision analysis, full spec history) that clearly belongs in separate reference files. References aren't just unclear — they point at files absent from the bundle.

2 / 5

Total

11

/

20

Passed

Description

58%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.

The description communicates a clear niche with concrete deliverables (CGD/SOT files) in proper third-person voice, but it reads as insider documentation rather than a trigger surface: it lacks any 'Use when...' guidance and leans on jargon ('epistemic quality', 'properly qualified') instead of the natural phrases users would say. The natural trigger terms exist in the skill's frontmatter but not in the description being evaluated.

Suggestions

Add an explicit 'Use when...' clause to the description (e.g., 'Use when ingesting documents into a RAG knowledge base, before sharing docs with other AI systems, or when a document mixes projections with facts') to raise completeness above the cap of 3.

Swap abstract phrasing like 'epistemic quality' and 'properly qualified' for concrete checks users recognize: flags unverified claims, missing uncertainty markers, and future-state-as-present statements that could cause hallucinations.

Include one or two natural user phrases (e.g., 'hallucination risk check', 'verify claims before RAG ingestion') in the description itself so trigger matching works on what users actually say, not just the frontmatter triggers list.

DimensionReasoningScore

Specificity

Names the domain (RAG pre-ingestion) and several concrete actions: 'Ensures documents are properly qualified', 'Produces CGD (Clarity-Gated Documents) and validates SOT (Source of Truth) files'. Not 5 because 'properly qualified' and 'epistemic quality' are abstract jargon that doesn't convey what checking actually happens; not 3 because four distinct actions are named with concrete output artifacts.

4 / 5

Completeness

The 'what' is clearly stated ('Pre-ingestion verification... Produces CGD... validates SOT'), but there is no 'Use when...' clause or equivalent explicit trigger guidance in the description, which caps completeness at 3 per the judging guidelines. 'Before entering knowledge bases' only weakly implies when to reach for the skill.

3 / 5

Trigger Term Quality

'Pre-ingestion', 'RAG systems', 'knowledge bases' are genuine domain keywords, but the natural phrases a user would say ('hallucination risk', 'unverified claims', 'hedge this') appear only in the separate triggers frontmatter, not the description itself. 'Epistemic quality', 'CGD', and 'SOT' are insider terms users rarely say aloud, so coverage misses common variations.

3 / 5

Distinctiveness Conflict Risk

A clear niche (epistemic gating for RAG ingestion) with distinct vocabulary (CGD, SOT, pre-ingestion) that is unlikely to fire for unrelated skills. Only minor overlap risk with generic document-review or fact-checking skills; not 5 because 'verification' and 'documents' are broad words that could collide with general review skills.

4 / 5

Total

14

/

20

Passed

Validation

81%

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

Validation — 13 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

skill_md_line_count

SKILL.md is long (714 lines); consider splitting into references/ and linking

Warning

frontmatter_unknown_keys

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

Warning

referenced_paths_exist

Referenced path issues: 8 missing

Warning

Total

13

/

16

Passed

Repository

administrakt0r/AI-Agents-Safe-Coding-Skills

Path

plugins/AI-Agents-Safe-Coding-Skills/skills/clarity-gate/SKILL.md

Revision

64adc55

Reviewed

Table of Contents

DiscoveryImplementationValidation

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.

Claim skill