exploit-command-injection

OS Command Injection — exploiting applications that pass user input to OS commands without sanitization. Covers injection operators (;, |, ||, &&, $(), backticks, newline), blind detection (time-based, OOB callback), and bypass techniques (space, keyword, encoding).

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Quality

Content

83%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.

A token-efficient, highly actionable command-injection playbook with copy-paste ready bash across detection, bypass, and exfiltration. Its weakness is workflow clarity around the batch credential-harvest step, which lacks an explicit validate/verify-then-retry feedback loop, capping that dimension.

Suggestions

Add an explicit validation/retry checkpoint after the credential-harvest loop: verify /tmp/rce_harvest.txt is non-empty and re-run with adjusted candidate paths if it comes back empty.

Sequence the workflow more explicitly with numbered steps (detect → confirm RCE → harvest → exfiltrate) and mark each checkpoint, so the multi-stage process reads as a controlled pipeline rather than parallel sections.

Add a brief verification step for the time-based oracle (e.g., confirm a known character returns correctly before trusting the full extraction loop) to close the feedback-loop gap on that batch operation.

Dimension	Reasoning	Score
Conciseness	The body is lean: copy-paste bash snippets with terse inline comments and no preamble explaining what command injection is or how shells work, so every token earns its place and assumes Claude's competence.	5 / 5
Actionability	It provides fully executable curl one-liners with placeholders across operators, blind detection, bypass, credential harvesting, and exfiltration, including an automated time-based oracle loop that covers the common cases copy-paste ready.	5 / 5
Workflow Clarity	A detect → confirm-RCE → harvest → exfiltrate sequence is present with one verification checkpoint ('Once you have confirmed RCE...'), but the batch credential-harvest loop has no validate/verify-then-retry feedback loop, and the rubric caps batch operations without validation at 3.	3 / 5
Progressive Disclosure	No bundle files exist; the single SKILL.md is organized into clear, well-signaled sections (Detection, Injection Operators, Bypass, Credential Extraction, Blind Exfiltration) with no nested references, giving good structure appropriate to a single-file skill, though it slightly exceeds the under-50-line simple-skill threshold.	4 / 5
	Total	17 / 20 Passed

Description

78%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.

A strong, specific description that names a clear niche and lists concrete capability categories with rich trigger terms in metadata. Its main gap is the absence of an explicit 'Use when...' clause inside the description prose itself, relying on metadata for trigger guidance.

Dimension	Reasoning	Score
Specificity	The description lists several concrete capability categories — injection operators, blind detection (time-based, OOB), and bypass techniques — rather than just naming the domain, but it enumerates categories rather than many distinct concrete actions, so it stops short of a 5.	4 / 5
Completeness	It clearly states what ('exploiting applications that pass user input to OS commands without sanitization' plus covered techniques) and provides equivalent explicit trigger guidance via when_to_use metadata, but the 'when' is not embedded as a 'Use when...' clause in the description prose itself, so it is not a 5.	4 / 5
Trigger Term Quality	The when_to_use metadata supplies strong natural trigger terms (cmdi, shell injection, popen, subprocess, shell=True, eval, passthru), giving good keyword coverage, though the description prose itself carries fewer natural synonyms, leaving a few terms implicit.	4 / 5
Distinctiveness Conflict Risk	'OS Command Injection' is a clearly defined niche with distinct, specific trigger terms that are unlikely to collide with unrelated skills, matching the clear-niche anchor.	5 / 5
	Total	17 / 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
metadata_version	'metadata.version' is missing	Warning

	Total	15 / 16 Passed

Repository: PurpleAILAB/Decepticon
Path: packages/decepticon/decepticon/skills/standard/exploit/web/command-injection/SKILL.md
Commit: 0cf691e

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.