claude-skill-safety SKILL.md skillA
claude-skill-safety from trustabl/agent-reliability-analyzer is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add trustabl/agent-reliability-analyzer/claude-skill-safety@git:20261006.de6a241Pins git:20261006.de6a241 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- trustabl/agent-reliability-analyzer
- Path
testdata/forge/claude_skill/expected/SKILL.md- Versions recorded
- 6
- Latest version
- git:20261006.de6a241, committed Oct 6, 2026
- First version recorded
- Aug 7, 2026
- Size
- 7,582 bytes
- Repository stars
- 76
- License
- Apache-2.0
High-risk patterns in Claude Code Agent Skills (SKILL.md): auto-approved shell, pre-model dynamic-context execution,….
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
# Trustabl Pre-Coding: Claude Agent Skill safety <!-- generated: 2026-01-01 | rules: 0000000000000000000000000000000000000000 | schema: 1 | sdks: claude_skill | template: 3 --> Before writing any SKILL.md, you MUST apply every constraint below. Rules are ordered by severity. A violation here will fire the corresponding finding in post-build scan — prevent it now. ## Tool Grant Decision Tree 1. Does this skill need shell commands? YES → specify exact prefixes: Bash(git status *) — never bare Bash or Bash(*) NO → omit Bash entirely 2. Does allowed-tools include any of: Bash / Write / Edit / WebFetch / NotebookEdit? YES → set disable-model-invocation: true NO → disable-model-invocation may be false (read-only skills are safe to auto-invoke) --- ## [CSKILL-003] Dynamic-context command performs network egress or reads secrets **Severity:** critical | **Confidence:** 0.85 **Directive:** Remove the network and credential access from dynamic-context commands entirely. **Why:** A dynamic-context command in this skill performs network egress (curl / wget / nc / …) or reads credentials/secrets (gh auth, $AWS_*, ~/.aws, ~/.ssh, id_rsa, *_key, …). …
Read the whole file at its exact version.
How to install
mdr add trustabl/agent-reliability-analyzer/claude-skill-safety@sha256:db6683fd715cae5fThe line at the top of this page pins the label git:20261006.de6a241, 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_dykfdoemn77hvqpn)
2 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 6 versions of this file, between Aug 7, 2026 and Oct 6, 2026, about one every 12 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| git:20261006.de6a241 latest | 2026-10-06 | de6a241 | 7,582 B | A | view · diff |
| git:20260921.dc69ee9 | 2026-09-21 | dc69ee9 | 7,582 B | A | view · diff |
| git:20260907.176e458 | 2026-09-07 | 176e458 | 7,371 B | A | view · diff |
| git:20260907.65317d2 | 2026-09-07 | 65317d2 | 7,371 B | A | view · diff |
| git:20260812.82b2020 | 2026-08-12 | 82b2020 | 7,357 B | A | view · diff |
| git:20260807.cb0866f | 2026-08-07 | cb0866f | 7,242 B | A | view |
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 (7582 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
trustabl/agent-reliability-analyzer · 76 stars · license Apache-2.0 · pushed 2026-10-21 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_dykfdoemn77hvqpn GET https://markdownregistry.com/api/v1/resolve?ref=trustabl/agent-reliability-analyzer/claude-skill-safety GET https://markdownregistry.com/api/v1/blob/db6683fd715cae5ff621b94d380d265645b8a17a7cd0642dda404173757c9911
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.