agent-refusal-with-evidence-beats-literal-compliance SKILL.md skillA
agent-refusal-with-evidence-beats-literal-compliance from wan-huiyan/agent-traffic-control is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add wan-huiyan/agent-traffic-control/agent-refusal-with-evidence-beats-literal-compliance@v1.0.0Pins v1.0.0 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- wan-huiyan/agent-traffic-control
- Path
plugins/agent-traffic-control/skills/agent-refusal-with-evidence-beats-literal-compliance/SKILL.md- Versions recorded
- 1
- Latest version
- v1.0.0, committed Sep 2, 2026
- Size
- 11,434 bytes
- Repository stars
- 3
- License
- MIT
A dispatched agent that does exactly what you asked can ship a defect you would have stopped yourself, because it can see something at the worksite that you could not see when you wrote the brief. Use when: (1) you are writing a brief that prescribes a specific fix, threshold, wording or value rather than the outcome you want; (2) an agent reports that it did not do what you asked and you are deciding whether that is insubordination or judgement; (3) an agent asserts "this is fine" or…
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
# Let agents refuse the letter of an instruction — and require the evidence ## Problem You write a brief. Somewhere in it you prescribe not just the outcome but the *method*: apply this fix, use this threshold, record this value, phrase it this way. By the time the agent gets to that line it is standing somewhere you were not when you wrote it. It has run the tests, read the surrounding code, seen the data. And sometimes what it can see is that **your instruction, carried out exactly, makes things worse**. A compliant agent does it anyway. That is the default, and it is a bad default: the agent's report comes back clean, the instruction was followed, and the defect ships with your name on the decision. **In a single run, four agents correctly declined what they were asked to do.** Not one of those was a failure. Every one was a defect caught at the last place it could still be caught for free. But refusal on its own is not the good outcome either. *"I didn't do that, it seemed wrong"* is unbankable: you cannot act on it, you cannot review it, and you have to re-derive the whole question yourself. What made these four useful was that each one …
Read the whole file at its exact version.
How to install
mdr add wan-huiyan/agent-traffic-control/agent-refusal-with-evidence-beats-literal-compliance@sha256:27dd784ef9d199feThe line at the top of this page pins the label v1.0.0, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_t2m4ejakuuko6yy7)
3 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 1 version of this file, committed on Sep 2, 2026.
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (11434 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
wan-huiyan/agent-traffic-control · 3 stars · license MIT · pushed 2026-09-22 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_t2m4ejakuuko6yy7 GET https://markdownregistry.com/api/v1/resolve?ref=wan-huiyan/agent-traffic-control/agent-refusal-with-evidence-beats-literal-compliance GET https://markdownregistry.com/api/v1/blob/27dd784ef9d199fe99c1ea25892fe71c43a19bdc331e97f232714b024c9fb731
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.