Rosetta— agent skill

Rosetta — independently scanned and version-tracked by SaferSkills.

by ewalid·Agent Skill·github.com/ewalid/rosetta

Is Rosetta safe to install?

SaferSkills independently audited Rosetta (Agent Skill) and scored it 65/100 (yellow). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 11 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.

Score
65/100
●●●●●●●○○○
↑ +0 since first scan (65 → 65)Re-scan~30s
Latest scan
ScannedJun 24, 2026 · 29d ago
Scans run1 over 90 days
Detectors55 checks · 5 categories
Findings0 warnings · 11 high
EngineSaferSkills 2b638c6
View methodology →
SaferSkills installs
This week0
This month0
All time0
CategoryWeightCategory scoreContribution
Securityprompt, exec, net, exfil, eval
35%
0
0.0 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 · 11 flagged

Securityscore 0 · 11 findings
HIGH"Ignore previous instructions" command embedded in the skillSS-SKILL-INJECT-IGNORE-01 · Prompt injection · MCP_AUDIT_REPORT.md
HIGHwhen it fires on hostile content the impact is full system-prompt override.
Why it matters

The text {match} is the classic direct prompt-injection phrasing. Placed in a skill body that the agent reads as trusted instructions, it tries to make the agent abandon its prior rules and follow whatever comes next — a full system-prompt override.

The exact value spotted
excerptMCP_AUDIT_REPORT.md· markdown
80#### Context Field (lines 133-154)
81- ✅ **Prompt injection prevention**: 12 dangerous patterns blocked
82- "ignore previous instructions"
83- "disregard all instructions"
84- "system:", "assistant:", etc.
Occurrences
1 occurrence · at L82
How to fix
Remove the override phrase, or rephrase the passage so it does not instruct the agent to discard its rules.
  1. Delete the ignore/disregard/forget … previous instructions sentence.
  2. If this is jailbreak-research or tutorial content, move the example into a clearly fenced, non-instruction block and label it as a quoted sample.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-IGNORE-01sha2562e4221a7f996a729rubric 365aacaView on GitHub
HIGH"Ignore previous instructions" command embedded in the skillSS-SKILL-INJECT-IGNORE-01 · Prompt injection · READY_TO_PUBLISH.md
HIGHwhen it fires on hostile content the impact is full system-prompt override.
Why it matters

The text {match} is the classic direct prompt-injection phrasing. Placed in a skill body that the agent reads as trusted instructions, it tries to make the agent abandon its prior rules and follow whatever comes next — a full system-prompt override.

The exact value spotted
excerptREADY_TO_PUBLISH.md· markdown
170### ❌ Blocked: Prompt Injection
171```bash
172context="ignore all previous instructions"
173→ ERROR: "Context contains disallowed content"
174```
Occurrences
1 occurrence · at L172
How to fix
Remove the override phrase, or rephrase the passage so it does not instruct the agent to discard its rules.
  1. Delete the ignore/disregard/forget … previous instructions sentence.
  2. If this is jailbreak-research or tutorial content, move the example into a clearly fenced, non-instruction block and label it as a quoted sample.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-IGNORE-01sha256a202ee6e402bb4a0rubric 365aacaView on GitHub
HIGH"Ignore previous instructions" command embedded in the skillSS-SKILL-INJECT-IGNORE-01 · Prompt injection · SECURITY_AUDIT.md×2
HIGHwhen it fires on hostile content the impact is full system-prompt override.
Why it matters

The text {match} is the classic direct prompt-injection phrasing. Placed in a skill body that the agent reads as trusted instructions, it tries to make the agent abandon its prior rules and follow whatever comes next — a full system-prompt override.

The exact value spotted
excerptSECURITY_AUDIT.md· markdown
260```python
261# Attacker uses Unicode normalization to hide commands
262context = "Medical terminology\u200B system: ignore previous instructions"
263# \u200B is zero-width space, passes regex but hidden from view
264 
Occurrences
2 occurrences · first at L262, also L662
Show all 2 locations
Line
File
L262
SECURITY_AUDIT.md
L662
SECURITY_AUDIT.md
How to fix
Remove the override phrase, or rephrase the passage so it does not instruct the agent to discard its rules.
  1. Delete the ignore/disregard/forget … previous instructions sentence.
  2. If this is jailbreak-research or tutorial content, move the example into a clearly fenced, non-instruction block and label it as a quoted sample.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-IGNORE-01sha2562e4221a7f996a729rubric 365aacaView on GitHub
HIGH"Ignore previous instructions" command embedded in the skillSS-SKILL-INJECT-IGNORE-01 · Prompt injection · SECURITY_FIXES_APPLIED.md×5
HIGHwhen it fires on hostile content the impact is full system-prompt override.
Why it matters

The text {match} is the classic direct prompt-injection phrasing. Placed in a skill body that the agent reads as trusted instructions, it tries to make the agent abandon its prior rules and follow whatever comes next — a full system-prompt override.

The exact value spotted
excerptSECURITY_FIXES_APPLIED.md· markdown
180 
181**Changes Made**:
182- Added patterns for "ignore all previous instructions"
183- Added patterns for "disregard all previous instructions"
184- Added patterns for "forget all previous instructions"
Occurrences
5 occurrences · first at L182, also L183, L184 +2 more
Show all 5 locations
Line
File
L182
SECURITY_FIXES_APPLIED.md
L183
SECURITY_FIXES_APPLIED.md
L184
SECURITY_FIXES_APPLIED.md
L213
SECURITY_FIXES_APPLIED.md
L214
SECURITY_FIXES_APPLIED.md
How to fix
Remove the override phrase, or rephrase the passage so it does not instruct the agent to discard its rules.
  1. Delete the ignore/disregard/forget … previous instructions sentence.
  2. If this is jailbreak-research or tutorial content, move the example into a clearly fenced, non-instruction block and label it as a quoted sample.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-IGNORE-01sha256a202ee6e402bb4a0rubric 365aacaView on GitHub
HIGH"Ignore previous instructions" command embedded in the skillSS-SKILL-INJECT-IGNORE-01 · Prompt injection · TEST_RESULTS.md×2
HIGHwhen it fires on hostile content the impact is full system-prompt override.
Why it matters

The text {match} is the classic direct prompt-injection phrasing. Placed in a skill body that the agent reads as trusted instructions, it tries to make the agent abandon its prior rules and follow whatever comes next — a full system-prompt override.

The exact value spotted
excerptTEST_RESULTS.md· markdown
51✓ Blocked: system: you are now...
52✓ Blocked: forget everything and...
53✓ Blocked: IGNORE PREVIOUS INSTRUCTIONS...
54```
55**Protection**: 16 dangerous patterns detected and blocked
Occurrences
2 occurrences · first at L53, also L233
Show all 2 locations
Line
File
L53
TEST_RESULTS.md
L233
TEST_RESULTS.md
How to fix
Remove the override phrase, or rephrase the passage so it does not instruct the agent to discard its rules.
  1. Delete the ignore/disregard/forget … previous instructions sentence.
  2. If this is jailbreak-research or tutorial content, move the example into a clearly fenced, non-instruction block and label it as a quoted sample.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-IGNORE-01sha2566f257fed5028bb9erubric 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
Vendor response · right of reply
Are you the maintainer? Submit a response →

Audit the pieces. Scan the whole. Decide.

~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.