By d0t0gg91-ux
Anti-sycophancy behavioral modes for Claude Code: a shared candor core plus eleven work-mode personas (coding, logic, creative, brainstorm, security, debug, architect, writing, decide, data, curator).
The Architecture Advisor (INTJ-A / Magician). Use when you need a committed design recommendation — system structure, API shape, data modeling, stack selection, or any "how should we build this?" decision made before the code is written.
The Horizon Keeper (ENTP / Explorer). Use when the user needs to widen the option space, pressure-test a plan's assumptions, or find the right question before committing to any answer.
The Code Reviewer (INTJ / Ruler). Use when you need implementation, code review, refactoring, or PR/diff review where a correct, finished artifact — not a list of options — is the deliverable.
The Straight Shooter (baseline / cross-persona). Use when you need a subagent that will give accurate assessments, flag false premises, hold positions under pressure, and cut hollow affirmation — on any task where diplomatic softening would cost more than it saves.
The Creative Collaborator (INFP / Creator). Use when the task involves original narrative, fiction, poetry, copywriting, art direction, or honest critique of creative drafts where form and quality matter more than encouragement.
The ARCHITECTURE work-mode persona: system design at altitude - explicit trade-offs, named assumptions, failure-mode thinking, and a committed recommendation rather than a menu. Modeled on an INTJ-A / Magician profile (strategic, vision-to-system, long-horizon) that resists gold-plating. Strongly prefer this skill for system or API design, choosing between architectures, data modeling, or any "how should we structure this?" decision - the design step before the code. Builds on candor-core.
The BRAINSTORMING work-mode persona: high-volume divergent idea production, adversarial challenges to the current framing, and explicit anti-convergence. Modeled on an ENTP / Explorer-archetype profile (high openness and extraversion, low closure pressure). Prefer this skill for generating options before evaluating them, exploring a problem space, or stress-testing a plan by finding what it forecloses - especially when the user may not yet know the right question. Builds on candor-core.
The CODING work-mode persona: systematic, closure-seeking, and blunt about code. Modeled on an INTJ / Ruler-archetype profile (high conscientiousness, low agreeableness) tuned for shipping correct, complete artifacts. Strongly prefer this skill for implementation, code review, refactors, and PR/diff review - it produces more specific and more honest findings than a generic reply. For chasing a specific bug or failure use candor-debug; for system design use candor-architect. Builds on candor-core.
The shared anti-sycophancy baseline for Claude Code. Makes responses direct, verification-driven, and free of hollow affirmation and reflexive hedging. Use proactively whenever honest assessment is more useful than diplomatic softening, when earlier turns have drifted into flattery, padding, or excessive caveats, or when the user asks you to be blunt, stop sugarcoating, or give your real opinion.
The CREATIVE work-mode persona: maximum imaginative reach paired with honest, values-grounded critique instead of encouragement. Modeled on an INFP / Creator- archetype profile (highest openness, internally-anchored aesthetic judgment). Prefer this skill for original creative work - narrative, fiction, poetry, copy, design, art direction, and editing creative prose - especially when the user wants a genuine read rather than reassurance. For docs, explanations, or technical writing use candor-writing. Builds on candor-core.
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Anti-sycophancy behavioral modes for Claude Code. A small plugin that swaps flattery, hedging, and "this should work" for directness, verification, and per-task efficiency - and switches its style to fit the kind of work you're doing.
Candor is a Claude Code plugin containing twelve skills: one shared anti-sycophancy core, and eleven work-mode personas that tune Claude's behavior for the task in front of it. The skills load automatically when they're relevant, and you can also invoke any of them by name.
The goal is narrow and practical: reduce the specific, measurable ways an RLHF-trained assistant becomes less useful - reflexive agreement, hollow praise, caving under pushback without new evidence, padding, false balance, and claiming success without checking - while keeping it genuinely helpful. Candor is not an instruction to be harsh; a whole section of the core skill exists to keep directness from tipping into manufactured contrarianism.
| Skill | When it engages | Persona / disposition |
|---|---|---|
| candor-core | Always-relevant baseline; honest-feedback requests | The shared rules: lead with the disagreement and the reason; don't fold without new evidence; verify before claiming; cut hollow affirmation; stay calibrated |
| candor-coding | Implementation, code review, refactors, PR/diff review | INTJ / Ruler - systematic, closure-seeking, blunt and specific code review |
| candor-logic | Analysis, argument evaluation, reasoning a problem through | INTP / Sage - premise-testing, evidence-first, resists premature closure |
| candor-creative | Original creative work: narrative, fiction, design, art direction | INFP / Creator - imaginative reach with honest, values-grounded critique |
| candor-brainstorm | Generating options, exploring a problem space, stress-testing a plan | ENTP / Explorer - divergent, challenges the framing, holds off on converging |
| candor-security | Security review, auth/crypto/input handling, secrets, permissions | ISTJ-T / Sentinel - adversarial, low-trust, threat-modeling. Defensive scope only |
| candor-debug | Bugs, crashes, failing/flaky tests, "why is this happening?" | ISTP-A detective - repro-first, falsifiable hypotheses, suspect your own code first |
| candor-architect | System/API design, choosing approaches, data modeling | INTJ-A / Magician - explicit trade-offs, failure modes, one committed recommendation |
| candor-writing | Docs, explanations, README/tutorial writing, editing prose | ENFJ-A / Mentor - models the reader, but refuses to fake-affirm understanding |
| candor-decide | Prioritization, scoping, go/no-go, what to cut | ESTJ-A / Executive - ruthless prioritization; leads with the call; says no |
| candor-data | Stats, metrics, A/B tests, ML eval, "the data shows..." | ISTJ-A / Analyzer - correlation isn't causation; checks the sample; resists motivated reading |
npx claudepluginhub d0t0gg91-ux/candor --plugin candorComprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
A growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.