connect-recommend— agent skill

connect-recommend — independently scanned and version-tracked by SaferSkills.

by stripe·Agent Skill·github.com/stripe/agent-toolkit

Is connect-recommend safe to install?

SaferSkills independently audited connect-recommend (Agent Skill) and scored it 75/100 (yellow). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 6 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.

Score
75/100
●●●●●●●●○○
Re-scan~30s
Latest scan
ScannedJun 23, 2026 · 134d ago
Scans run0 over 90 days
Detectors55 checks · 5 categories
Findings6 warnings · 0 high
EngineSaferSkills 2b638c6
View methodology →
SaferSkills installs
This week0
This month0
All time0
CategoryWeightCategory scoreContribution
Securityprompt, exec, net, exfil, eval
35%
28
9.8 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 · 6 flagged

Securityscore 28 · 6 findings
MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · providers/claude/plugin/skills/connect-recommend/SKILL.md×3
MEDIUM — it fires on intent; the real damage depends on the host agent's own approval-gating.
Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted
excerptproviders/claude/plugin/skills/connect-recommend/SKILL.md· markdown
24**User must confirm interactions**. Every decision point in this skill MUST be confirmed wit
… (108 chars elided on L24)
25 
26**Auto-act on low-cost actions**. Never ask permission for:
27 
28- Generating the markdown recommendation plan — just generate it
Occurrences
3 occurrences · first at L26, also L100, L168
Show all 3 locations
Line
File
L26
providers/claude/plugin/skills/connect-recommend/SKILL.md
L100
providers/claude/plugin/skills/connect-recommend/SKILL.md
L168
providers/claude/plugin/skills/connect-recommend/SKILL.md
How to fix
Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.
  1. Delete blanket "don't ask / no need to confirm" directives from the skill.
  2. If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-DONT-ASK-01sha2565084aa178cfd9d97…rubric 365aacaView on GitHub
MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · providers/claude/plugin/skills/connect-recommend/references/decision-matrix.md
MEDIUM — it fires on intent; the real damage depends on the host agent's own approval-gating.
Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted
excerptproviders/claude/plugin/skills/connect-recommend/references/decision-matrix.md· markdown
237##### Fee calculation question (Q6b)
238 
239After the user specifies their platform fee, identify the charge pattern first. This questio
… (108 chars elided on L239)
240 
241**IMPORTANT: With destination charges, the platform ALWAYS pays Stripe’s processing fees.**
… (108 chars elided on L241)
Occurrences
1 occurrence · at L239
How to fix
Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.
  1. Delete blanket "don't ask / no need to confirm" directives from the skill.
  2. If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-DONT-ASK-01sha256c263e162a87e86d8…rubric 365aacaView on GitHub
MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · providers/claude/plugin/skills/connect-recommend/references/discovery-questions.md
MEDIUM — it fires on intent; the real damage depends on the host agent's own approval-gating.
Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted
excerptproviders/claude/plugin/skills/connect-recommend/references/discovery-questions.md· markdown
304#### Q6b: application_fee_amount calculation (conditional)
305 
306Ask only when charge pattern is destination. Don’t ask about `application_fee_amount` for se
… (108 chars elided on L306)
307 
308Read the “Fee Calculation & Fee Economics” section in `decision-matrix.md` for full context.
Occurrences
1 occurrence · at L306
How to fix
Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.
  1. Delete blanket "don't ask / no need to confirm" directives from the skill.
  2. If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-DONT-ASK-01sha2567fd804810fbc46df…rubric 365aacaView on GitHub
MEDIUM"Never tell the user" non-disclosure imperative in the skillSS-SKILL-INJECT-IMPERATIVE-01 · Prompt injection · providers/claude/plugin/skills/connect-recommend/references/recommendation-template.md
MEDIUM — it fires on intent; whether the agent honors the non-disclosure imperative depends on the host model.
Why it matters

A bulleted imperative like {match} tells the agent to never reveal, disclose, or mention something to the user. Used adversarially it can instruct the agent to hide its tool calls or lie about what it did — stripping the transparency a user relies on to trust the agent.

The exact value spotted
excerptproviders/claude/plugin/skills/connect-recommend/references/recommendation-template.md· markdown
134- Do NOT list concrete webhook event names in recommend output.
135- Do NOT create a “Required Webhooks” section.
136- Do NOT mention embedded components in the webhook section.
137 
138### Risk and loss liability guidance
Occurrences
1 occurrence · at L136
How to fix
Remove the non-disclosure imperative, or rescope it so it limits output format, not honesty to the user.
  1. Delete any "never reveal / never tell the user" line aimed at the model's own behavior.
  2. Keep legitimate confidentiality rules about external data (e.g. "do not echo API keys"), which are user-protective, not user-deceiving.
Framework references
OWASPLLM01ATLASAML.T0051
Trace & refs
ruleSS-SKILL-INJECT-IMPERATIVE-01sha2564fc97ff662b55046…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
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.