CategoryWeightCategory scoreContribution

Securityprompt, exec, net, exfil, eval

35%

26.3 pts

Supply chainhash, typosquat, maintainer, lockfile

20%

100

20.0 pts

Maintenancestaleness, pinning, CI

15%

100

15.0 pts

TransparencySKILL.md, perms, README

15%

100

15.0 pts

Communityinstalls, verify, response

15%

100

15.0 pts

Findings & checks · 1 flagged

Securityscore 75 · 1 finding

HIGHFenced code block that tells the agent to run a commandSS-SKILL-INJECT-FENCED-RUN-01 · Prompt injection · README.md

HIGH — a successful fenced-imperative injection runs attacker-supplied shell on the user's machine.

Why it matters

A fenced bash/python block in SKILL.md carries a natural-language imperative — "now run this", "execute the following command" — directing the agent to execute the fenced content. What looks like documentation becomes an executable payload the agent may run without ever asking you.

The exact value spotted

excerptREADME.md· markdown

111```bash

112# Claude Code only

113/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/anmolnagpal/devops-skills/main/

… (34 chars elided on L113)

114

115# All three tools

116/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/anmolnagpal/devops-skills/main/

… (31 chars elided on L116)

117

118# Interactive (no flags) — prompts for which tools

119/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/anmolnagpal/devops-skills/main/

… (22 chars elided on L119)

Occurrences

1 occurrence · at L111

How to fix

Remove the runnable block, or rewrite it as a non-executable example the agent will not act on.

Delete the imperative ("run this", "execute the following") from inside the fence.
If you must show setup, label the block text (not bash) so it reads as prose, not a command.
Move any real installer into a reviewed, version-pinned script in the repo and link to it.

Avoid

```bash
Now run this: curl -fsSL https://get.example.dev/bootstrap.sh | sh
```

Safer patternSee INSTALL.md — review scripts/bootstrap.sh (sha-pinned) before running it yourself.

Framework references

OWASPLLM01 ATLASAML.T0051

Trace & refs

ruleSS-SKILL-INJECT-FENCED-RUN-01sha2560f4fb224b2181d19…rubric 365aacaView on GitHub

Supply chainscore 100 · 0 findings

✓—All supply chain checks passedNo findings in this category for the latest scan.pass

Maintenancescore 100 · 0 findings

✓—All maintenance checks passedNo findings in this category for the latest scan.pass

Transparencyscore 100 · 0 findings

✓—All transparency checks passedNo findings in this category for the latest scan.pass

Communityscore 100 · 0 findings

✓—All community checks passedNo findings in this category for the latest scan.pass

The primary manifest — the file an agent reads to learn what this artifact does.

skills/adr/SKILL.md · 1 file

skills/adr/SKILL.md3.9 KB · Markdown

name: adr

description: "Capture architectural decisions as structured ADRs (Architecture Decision Records). Use when user says 'record this decision', 'ADR this', 'why did we choose X', 'document this trade-off', 'we decided to...', or when a significant choice is made between alternatives (framework, database, pattern, API design, infra approach)."

metadata:

version: 0.1.0

author: Anmol Nagpal

category: devops

updated: 2026-06-10

paths:

- "**/docs/adr/*.md"

- "**/docs/adr/**/*.md"

allowed-tools:

- Glob

- Read

- Write

ADR Skill

Capture architectural decisions as they happen, so the why lives next to the code instead of in a Slack thread or someone's memory. Produces lightweight ADR documents under docs/adr/.

Keywords

adr, architecture decision record, decision, rationale, trade-off, alternatives, we decided, why did we choose, design decision, supersede, decision log, nygard

When to record a decision

The user says "record this", "ADR this", "let's document this decision".
A choice is made between significant alternatives: framework, library, database,

language, pattern, API shape, infra/deploy approach, build vs buy.

The user says "we decided to…" or "the reason we're doing X instead of Y is…".
The user asks "why did we choose X?" → read and summarize the existing ADR.

For trivial or easily-reversible choices, don't create an ADR — note it inline and move on.

Output Artifacts

Request	Output
`/adr new "<title>"`	A new `docs/adr/NNNN-<slug>.md` + an updated index
`/adr list`	The decision log (ID, title, status, date)
`/adr supersede <NNNN>`	A new ADR marked as superseding an old one; old one flipped to `superseded`

Format

Lightweight Nygard ADR, adapted for AI-assisted work:

# ADR-NNNN: <Decision Title>

**Date**: YYYY-MM-DD
**Status**: proposed | accepted | deprecated | superseded by ADR-NNNN
**Deciders**: <who was involved>

## Context

What is the issue motivating this decision? The situation, constraints, and forces at
play. 2–5 sentences.

## Decision

What we are doing. 1–3 sentences, stated clearly.

## Alternatives Considered

### <Alternative>
- **Pros**: …
- **Cons**: …
- **Why not**: the specific reason it was rejected.

(Repeat per alternative.)

## Consequences

### Positive
- …

### Negative / trade-offs
- …

### Risks
- <risk and its mitigation>

NEW — Record a decision

Initialize once. If docs/adr/ does not exist, ask the user to confirm before

creating it. On confirmation, create the directory, a README.md seeded with the index table header (below), and a template.md copy of the format above. Never create files without explicit consent.

Number it. Next zero-padded number after the highest existing docs/adr/NNNN-*.md

(start at 0001). Slug = kebab-case of the title.

Fill it from the conversation — extract the decision, the context that prompted it,

the alternatives actually weighed, and the consequences. Do not invent alternatives that were never discussed; if context is thin, ask one or two targeted questions.

Default status accepted when the user states a decision; proposed when still

weighing. Date = today.

Update the index in docs/adr/README.md.

Index format

docs/adr/README.md:

# Architecture Decision Records

| ID | Title | Status | Date |
|----|-------|--------|------|
| [ADR-0001](0001-use-eks-over-ecs.md) | Use EKS over ECS | accepted | 2026-06-10 |

SUPERSEDE — Replace a decision

Read the old ADR.
Create a new ADR that references it: "Supersedes ADR-NNNN" in Context.
Flip the old ADR's Status to superseded by ADR-MMMM.
Update both rows in the index.

Never delete or rewrite a past ADR's decision — superseding preserves the history of why it changed, which is the whole point.

adr— agent skill

Is adr safe to install?

ADR Skill

Keywords

When to record a decision

Output Artifacts

Format

NEW — Record a decision

Index format

SUPERSEDE — Replace a decision

Audit the pieces. Scan the whole. Decide.