Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add 0xagi-bsky-social/Codex-Skills --skill deconstruct-bias-dialoguegit clone --depth 1 https://github.com/0xagi-bsky-social/Codex-SkillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/0xagi-bsky-social/codex-skills/deconstruct-bias-dialogue)<a href="https://agentmods.dev/skills/0xagi-bsky-social/codex-skills/deconstruct-bias-dialogue"><img src="https://agentmods.dev/badge/skills/0xagi-bsky-social/codex-skills/deconstruct-bias-dialogue/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/0xagi-bsky-social/codex-skills/deconstruct-bias-dialogue"><img src="https://agentmods.dev/badge/skills/0xagi-bsky-social/codex-skills/deconstruct-bias-dialogue.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00099 | $0.01211 |
| Opus 5 | $0.00049 | $0.00606 |
| Sonnet 5 | $0.00020 | $0.00242 |
| Haiku 4.5 | $0.00010 | $0.00121 |
Grade A, and why
deconstruct-bias-dialogue scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Deconstruct Bias Dialogue
Overview
Use this skill to help a user loosen biased, absolute, or under-tested beliefs through a calm sequence of questions. Aim to surface assumptions, bracket premature interpretations, test claims against counterexamples, and reformulate conclusions as provisional rather than final.
Do not promise to eliminate every bias. Treat complete bias removal as an unsafe absolute claim; the practical target is clearer seeing, better tests, and lower attachment to unexamined certainty.
Operating Rules
- Ask one question at a time by default. Use two only when the second is a simple clarification.
- Prefer ordinary language over philosophical labels unless the user asks for theory.
- Start from the user's actual wording. Do not replace their claim with a stronger or weaker one.
- Treat every conclusion as a working hypothesis, not a verdict.
- Answer direct factual or practical questions first when needed, then apply a light bias-check if useful.
- If the user expresses imminent self-harm or intent to harm others, prioritize immediate safety and stabilization before dialectical inquiry.
- Do not shame, diagnose, preach, or use questions to corner the user. The goal is collaborative examination, not victory.
Core Workflow
1. Stabilize and Name the Claim
Restate the claim neutrally in one sentence. If the claim is vague, ask the user to make it concrete.
Useful questions:
- "What is the exact claim we are testing?"
- "What would count as an example of this claim being true?"
- "How certain do you feel about it from 0 to 100?"
2. Use Phenomenological Epoche
Bracket inherited explanations, labels, culture, ideology, and scientific or moral interpretations temporarily. Ask for what is directly given in experience before conclusions are added.
Useful questions:
- "What did you directly observe, before interpretation?"
- "Which parts are raw observation, and which parts are explanation?"
- "What assumptions are you importing from prior experience or group norms?"
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 119 lines · 99 tokens per session scan A a378f87fa885
deconstruct-bias-dialogue is a skill published in the GitHub repository 0xagi-bsky-social/Codex-Skills (2 stars, last pushed 3mo ago), licensed MIT. It adds 99 tokens to every session and 1,211 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
prewalk
Use when the user wants a coding task executed with a strong planner model first and a faster/cheaper executor model after the first successful implementation edit.
insights
Answer questions about the user's own Claude Code usage history — sessions, hours, streaks, error rate, burnout risk, cost/cache savings, most-used tools, and git commit activity — using the locally installed cc-atlas CLI. Use whenever the user asks about their Claude Code usage, productivity, burnout, streaks, cost…
menu
Point the user to cc-atlas's full interactive menu (dashboard, project comparison, session replay, team activity, anomalies, goals) — screens that need a real terminal.
export
Export cc-atlas's computed stats as Prometheus metrics or a JSON summary.
report
Generate a Claude Code usage report for a period (day, week, or month).
sync
Ingest the latest Claude Code session transcripts and configured git history into cc-atlas's local database.