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 EvoClaw/amplify --skill anti-cherry-pickgit clone --depth 1 https://github.com/EvoClaw/amplifyWrote 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/evoclaw/amplify/anti-cherry-pick)<a href="https://agentmods.dev/skills/evoclaw/amplify/anti-cherry-pick"><img src="https://agentmods.dev/badge/skills/evoclaw/amplify/anti-cherry-pick.svg" alt="Measured on agentmods" 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 | $0.00044 | $0.00722 |
| Opus 5 | $0.00022 | $0.00361 |
| Sonnet 5 | $0.00009 | $0.00144 |
| Haiku 4.5 | $0.00004 | $0.00072 |
Grade A, and why
anti-cherry-pick 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 8d 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 — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Anti-Cherry-Pick (Discipline Layer)
Overview
Selective reporting is not curation — it is fabrication. This skill activates at Phase 4 start and remains active until project end. Every experimental result, favorable or not, is recorded and reported.
Violating the letter of this rule is violating the spirit of this rule.
The Iron Law
NO SELECTIVE REPORTING. ALL SEEDS, ALL RESULTS, ALL FAILURES.
Mandatory Reporting Rules
-
All seeds, always. Run ALL pre-defined seeds from
evaluation-protocol.yaml. Report mean ± std or mean [95% CI]. Never report a single seed. -
Negative results are data. Failed experiments and negative results go in
docs/05_execution/negative-results.md. They cannot be silently dropped, deleted, or omitted from the final report. -
Fair baselines. Baselines must use fair configurations: same compute budget, same data, same preprocessing pipeline, same hyperparameter search budget. No handicapping.
-
Equivalent access. If your method uses pretrained models, extra data, or external resources, baselines must have equivalent access. Asymmetric advantage is not a fair comparison.
-
All datasets. If a dataset is listed in
evaluation-protocol.yaml, report results on it. You cannot include only favorable datasets and hide unfavorable ones. -
Equal tuning. The same hyperparameter search budget applies to all methods. You cannot exhaustively tune your method while giving baselines default parameters.
Red Flags — STOP
- Reporting only the best seed out of multiple runs
- Using different hyperparameter search budgets for own method vs. baselines
- Silently dropping a dataset where results are unfavorable
- Giving baselines a weaker configuration than your method
- Omitting a failed experiment from the record
- Reporting results before all seeds have completed
Rationalization Prevention
| Excuse | Reality |
|---|---|
| "This seed gave bad results" | Report ALL seeds. One bad seed in five is expected variance, not grounds for exclusion. |
| "The baseline code didn't work well" | Your implementation may be unfair. Use official code or explicitly acknowledge the discrepancy. |
| "This dataset isn't standard" | If you included it in evaluation-protocol.yaml, you report it. Remove it only through the change request process. |
| "Our method needs more tuning" | Same tuning budget for all methods. Equal effort, equal opportunity. |
| "The failure was due to a bug" | Fix the bug, re-run ALL experiments from scratch, report the new results. Partial re-runs are not acceptable. |
| "Negative results aren't interesting" | They are data. Record them. They prevent others from repeating your mistakes. |
| "This variant isn't worth reporting" | If you ran it, you report it. Let the reader decide what's worth noting. |
| "We'll include it in the appendix" | Appendix is acceptable for space. Omission is not. Ensure it is referenced from the main text. |
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.
- 8d ago First seen · 64 lines · 44 tokens per session scan A eb446f9aa2c6
anti-cherry-pick is a skill published in the GitHub repository EvoClaw/amplify (12 stars, last pushed 6mo ago), licensed MIT. It adds 44 tokens to every session and 722 once invoked, about $0.0002 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 2026-08-30.
Other skills, from other repositories
dingtalk_channel_connect
Use a headed browser to automatically complete DingTalk channel integration for QwenPaw. Applicable when the user mentions DingTalk, developer console, Client ID, Client Secret, bot, Stream mode, binding or configuring a channel. Supports pausing when a login page is detected and resuming after the user logs in.
A set of instructions for working with PDF files, which are documents designed to preserve their layout across devices.
make_plan
For external plan request scenarios, guides the Agent to request a clear, actionable, step-by-step plan from a stronger Agent via listagents and chatwithagent, emphasizing that the plan is executed by the requester, not by the consulted Agent.
officecli-word-form
Use this skill to create fillable Word forms (.docx) with real Content Controls (SDT) + legacy FormField checkboxes + MERGEFIELD mail-merge placeholders + document protection. Trigger on: 'fillable form', 'form fields', 'content controls', 'SDT', 'word form', 'fill in', 'only editable fields', 'protect document'…
officecli-data-dashboard
Use this skill to build a multi-element Excel dashboard — Dashboard sheet on open, multiple formula-driven KPI cards, multiple charts, sparklines, and conditional formatting — from CSV or tabular input. Trigger on: 'dashboard', 'KPI dashboard', 'analytics dashboard', 'executive dashboard', 'metrics dashboard', 'CSV to…
gpt-image-2
A skill for generating or editing images with GPT Image 2 across local, host-provided, or advisory setups.