anti-sycophancy

Eliminate sycophantic agreement patterns in AI responses. Challenge user claims independently, avoid agreement bias, and state evidence before deference.

Quality

69%

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

Fix and improve this skill with Tessl

tessl review fix ./skills/anti-sycophancy/SKILL.md

Quality

Content

85%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 well-structured behavioral skill that provides clear, actionable steps for avoiding sycophantic patterns. The workflow is unambiguous and appropriately scoped. The main gap is the absence of concrete before/after examples showing what a sycophantic vs. non-sycophantic response looks like, which would make the guidance more immediately actionable.

Suggestions

Add 1-2 concrete before/after examples showing a sycophantic response vs. the corrected response following the 4-step process, to make the behavioral change more tangible.

Dimension	Reasoning	Score
Conciseness	The content is efficient and avoids explaining what sycophancy is or why it's bad—concepts Claude already knows. Minor verbosity in the 'Limitations' section (e.g., 'This skill changes response posture, not factual access' is somewhat obvious), but overall tight.	4 / 5
Actionability	The 4-step process and the 3-step pushback handling are concrete and specific behavioral instructions. However, this is an instruction-only skill with no examples of input/output showing what a non-sycophantic response looks like versus a sycophantic one, which would strengthen actionability.	4 / 5
Workflow Clarity	This is a simple behavioral skill with a clear, unambiguous sequence: extract → assess → conclude → respond. The pushback handling adds a well-defined branching workflow. No destructive or batch operations are involved, so no validation checkpoints are needed.	5 / 5
Progressive Disclosure	This is a short, single-purpose skill under 50 lines with no need for external references. The content is well-organized into clear sections (When to Use, Process, Limitations) and appropriately self-contained.	5 / 5
	Total	18 / 20 Passed

Description

53%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 identifies a clear and relatively niche domain—anti-sycophancy in AI responses—and lists several behavioral directives. However, it lacks an explicit 'Use when...' clause, which is critical for skill selection, and the trigger terms lean toward technical AI jargon rather than natural user language. The actions described are more like behavioral principles than concrete, actionable steps.

Suggestions

Add an explicit 'Use when...' clause with natural trigger phrases, e.g., 'Use when the user asks Claude to be more critical, push back, give honest feedback, or stop being agreeable.'

Include natural-language synonyms and phrases users might say, such as 'stop agreeing with everything', 'be more honest', 'give me real feedback', or 'don't sugarcoat'.

Make the actions more concrete by specifying observable behaviors, e.g., 'Presents counterarguments before agreeing, flags unsupported claims, and provides balanced assessments even when they contradict the user's position.'

Dimension	Reasoning	Score
Specificity	Names the domain (sycophantic agreement patterns) and lists a few concrete actions ('challenge user claims independently', 'avoid agreement bias', 'state evidence before deference'), but these are somewhat abstract behavioral guidelines rather than truly concrete, comprehensive actions.	3 / 5
Completeness	Has a clear 'what' (eliminate sycophantic agreement patterns, challenge claims, avoid bias) but lacks an explicit 'when' clause. There is no 'Use when...' guidance to tell Claude when to select this skill, which caps this dimension at 3 per the rubric guidelines.	3 / 5
Trigger Term Quality	Includes relevant terms like 'sycophantic', 'agreement', 'bias', and 'deference', but misses natural user phrases like 'stop agreeing with me', 'be more critical', 'push back', 'honest feedback', or 'don't just tell me what I want to hear'.	3 / 5
Distinctiveness Conflict Risk	The focus on sycophancy and agreement bias is a fairly distinct niche that is unlikely to overlap with most other skills. However, it could potentially overlap with general 'tone' or 'communication style' skills, creating minor conflict risk.	4 / 5
	Total	13 / 20 Passed

Validation

90%

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 — 10 / 11 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	10 / 11 Passed

Repository: administrakt0r/AI-Agents-Safe-Coding-Skills
Path: skills/anti-sycophancy/SKILL.md
Commit: 3b73fbb

Reviewed: ~15 hours 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.