dangerous-actions

Rosetta CRITICAL MUST skill. MUST activate when action or its consequence is potentially dangerous, potentially irreversible, potentially destructive, or HIGH RISK. MUST activate when consequence MAYBE dangerous even if action itself seems safe. This is enterprise environment — the cost of dangerous activities is EXTREMELY HIGH, recovery may be impossible, and blast radius may affect production, shared environments, or other teams. If there is even a remote chance - load the skill.

Quality

56%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./instructions/r2/core/skills/dangerous-actions/SKILL.md

Quality

Content

68%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is admirably concise and well-sectioned for a simple gating skill, but its process steps are abstract and reference an undefined "hard-deny tier", and it lacks the validation/feedback loop expected of a destructive-operations workflow.

Suggestions

Define the "hard-deny tier" referenced in step 4 — list the concrete action classes that require explicit approval (it is currently a dangling "see below" with no target).

Add an explicit validation/verification checkpoint, e.g. a blast-radius checklist and a confirm-before-proceeding loop, so the destructive workflow has a feedback step.

Convert the <process>/<pitfalls> XML tags to standard markdown headers (## Process, ## Pitfalls) for clearer navigation.

Dimension	Reasoning	Score
Conciseness	The body is lean — short bullets and directives with no padding or explanation of concepts Claude already knows; every line (assess blast radius, think the opposite, require approval, pitfalls) earns its place.	5 / 5
Actionability	The examples list is concrete, but the core process steps are high-level directives ("Assess BLAST RADIUS", "THINK THE OPPOSITE") and step 4 references a "hard-deny tier (see below)" that is never defined — a missing key detail.	3 / 5
Workflow Clarity	A rough sequence exists, but this destructive-operations skill has no validation/verification checkpoint or feedback loop, and exceptions (numbered 5–7) interrupt the flow; the rubric caps destructive skills without validation at 3.	3 / 5
Progressive Disclosure	It is a short (<50 line) single-file skill organized into <process> and <pitfalls> sections with no need for external references, but uses non-standard XML tags instead of markdown headers and leaves a dangling "see below" reference.	4 / 5
	Total	15 / 20 Passed

Description

45%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is a strong, explicit trigger/activation spec but fails to state what the skill actually does, leaning heavily on MUST-activate command tone rather than concrete capability description. Its breadth ("even a remote chance") raises conflict risk.

Suggestions

Add a concrete "what" clause stating the skill's actions, e.g. "Assesses blast radius of risky actions, requires explicit user approval, and flags destructive operations before execution."

Replace some MUST-activate command phrasing with natural user-language triggers like "Use when an action could delete data, force-push, drop tables, or modify shared infrastructure."

Tighten the "even a remote chance - load the skill" breadth to reduce overlap with routine skills.

Dimension	Reasoning	Score
Specificity	The description names the domain ("potentially dangerous, potentially irreversible, potentially destructive, or HIGH RISK") but never states concrete actions the skill performs — it is entirely trigger language with no "what" capability listed.	2 / 5
Completeness	It explicitly and thoroughly answers "when" ("MUST activate when...") but the "what" is absent — no capability is stated — so it sits between the "only when present" (2) and "both what and when" (4) anchors; the strong explicit trigger guidance lifts it to the midpoint.	3 / 5
Trigger Term Quality	It includes relevant natural danger terms ("dangerous", "destructive", "irreversible", "high risk") but phrases them as MUST-activate commands rather than natural user speech, and misses common variations users actually say (e.g. "delete", "force-push", "drop table").	3 / 5
Distinctiveness Conflict Risk	The dangerous/irreversible/destructive niche is somewhat distinct, but "If there is even a remote chance - load the skill" is so broad it could overlap with many skills touching production, git, databases, or shared infrastructure.	3 / 5
	Total	11 / 20 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: griddynamics/rosetta
Path: instructions/r2/core/skills/dangerous-actions/SKILL.md
Commit: 70d401e

Reviewed: 2 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.