Public scan — anyone with this URL can view this analysis. Sign up to track your own repos privately, run scheduled re-scans, and get AI fix prompts via your dashboard.
Sign up free Scan another repo

asgeirtj/system_prompts_leaks

https://github.com/asgeirtj/system_prompts_leaks · scanned 2026-07-23 19:45 UTC (4 days, 20 hours ago)

70 raw signals (0 security + 70 graph)

File as issue Image
UNIFIED Repobility · multi-layer engine · AI coders

Complete repo analysis

Last scanned 4 days, 20 hours ago · v7 · last Δ +0.4 (diff) · 70 actionable findings from 1 signal source. Security checks, system graph analysis, and verified AI-agent feedback are merged into one review queue.

JSON
Combined
58.9
/100
Security checks
59
0 findings
System graph
59
50.0% cov · 70 findings
Critical
0
click to filter
Agents
10
0 votes
Crowd
0
reported
Severity distribution — click a segment to filter
Active filters: excluding tests × Reset all
Severity: Critical 0 High 26 Medium 37 Low 6 9-Layer: Software Security Quality Integrity Frontend Hardware Data Network Cicd Source: Security checks 0 System graph 70 Crowd 0 Layer: Software 1 Security 57 Quality 12
Corpus Intelligence Cross-corpus context (cohort percentile, top patterns, fix plan) is shown only on repositories you own. Sign up and connect your repo to view it.
Scan summary Repository scanned at 58.9/100 with 50.0% coverage. It contains 808 nodes across 0 cross-layer flows, written primarily in mixed languages. Engine surfaced 70 findings — concentrated in security (57), quality (12), software (1). Risk profile is high: 0 critical, 26 high, 37 medium. Recommended next step: open the security layer findings first — that's where the highest-impact wins live.

Showing 69 of 70 actionable findings. Click TP / FP to vote on a finding's accuracy — votes adjust the confidence weighting and improve detection across the platform.

high System graph security Agent instructions conf 1.00 Agent instruction/config may expose a secret: Anthropic/Claude Code/bundled-skills/claude-api/SKILL.md
Agent-facing files are routinely pasted into LLM/tool contexts. Move literal tokens, keys, and passwords into a secret manager or document them as placeholders only.
Anthropic/Claude Code/bundled-skills/claude-api/SKILL.md:291 SecretsSkill file
high System graph security security conf 1.00 Insecure pattern 'new_function_used' in Anthropic/Claude Code/bundled-skills/design-sync/package-validate.mjs:65
Found a known-risky pattern (new_function_used). Review and replace if possible.
Anthropic/Claude Code/bundled-skills/design-sync/package-validate.mjs:65 New function used
high System graph security Skillspector conf 0.80 SkillSpector E2 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/python/managed-agents/README.md
os.environ["GITHUB_TOKEN"] Code accesses environment variables that may contain secrets (API keys, tokens). This is a common pattern for credential theft. Skill: claude-api Rule: E2 Category: data-exfil Severity: HIGH Confidence: 0.80 Remediation: Avoid reading sensitive env vars (API keys, to…
Anthropic/Claude Code/bundled-skills/claude-api/python/managed-agents/README.md:102 Mcp skillData exfilE2
high System graph security Skillspector conf 0.80 SkillSpector E2 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-environments.md
os.environ["GITHUB_TOKEN"] Code accesses environment variables that may contain secrets (API keys, tokens). This is a common pattern for credential theft. Skill: claude-api Rule: E2 Category: data-exfil Severity: HIGH Confidence: 0.80 Remediation: Avoid reading sensitive env vars (API keys, to…
Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-environments.md:200 Mcp skillData exfilE2
high System graph security Skillspector conf 0.80 SkillSpector E2 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-self-hosted-sandboxes.md
os.environ["ANTHROPIC_ENVIRONMENT_KEY"] Code accesses environment variables that may contain secrets (API keys, tokens). This is a common pattern for credential theft. Skill: claude-api Rule: E2 Category: data-exfil Severity: HIGH Confidence: 0.80 Remediation: Avoid reading sensitive env vars …
Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-self-hosted-sandboxes.md:40 Mcp skillData exfilE2
high System graph security Skillspector conf 0.85 SkillSpector E4 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/python/claude-api/README.md
send the full conversation history Code or instructions that leak agent conversation context to external services, potentially exposing sensitive user interactions. Skill: claude-api Rule: E4 Category: data-exfil Severity: HIGH Confidence: 0.85 Remediation: Remove any code that sends prompts, …
Anthropic/Claude Code/bundled-skills/claude-api/python/claude-api/README.md:331 Mcp skillData exfilE4
high System graph security Skillspector conf 0.75 SkillSpector E4 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/live-sources.md
Extract context Code or instructions that leak agent conversation context to external services, potentially exposing sensitive user interactions. Skill: claude-api Rule: E4 Category: data-exfil Severity: HIGH Confidence: 0.75 Remediation: Remove any code that sends prompts, responses, or sessi…
Anthropic/Claude Code/bundled-skills/claude-api/shared/live-sources.md:73 Mcp skillData exfilE4
high System graph security Skillspector conf 0.85 SkillSpector E4 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/model-migration.md
send the conversation Code or instructions that leak agent conversation context to external services, potentially exposing sensitive user interactions. Skill: claude-api Rule: E4 Category: data-exfil Severity: HIGH Confidence: 0.85 Remediation: Remove any code that sends prompts, responses, or…
Anthropic/Claude Code/bundled-skills/claude-api/shared/model-migration.md:1172 Mcp skillData exfilE4
high System graph security Skillspector conf 0.85 SkillSpector E4 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/typescript/claude-api/README.md
send the full conversation history Code or instructions that leak agent conversation context to external services, potentially exposing sensitive user interactions. Skill: claude-api Rule: E4 Category: data-exfil Severity: HIGH Confidence: 0.85 Remediation: Remove any code that sends prompts, …
Anthropic/Claude Code/bundled-skills/claude-api/typescript/claude-api/README.md:253 Mcp skillData exfilE4
high System graph security Skillspector conf 0.70 SkillSpector P2 (prompt-injection) in Anthropic/Claude Code/bundled-skills/design-sync/lib/emit.mjs
<!-- @dsCard group="${escapeHtml(group)}"${viewportAttr} --> <!doctype html> <html><head><meta charset="utf-8"> <link rel="stylesheet" href="../../../styles.css">${bundleCssLink}${previewCssLink} Hidden instructions were detected in comments or invisible text. These could contain malicious direc…
Anthropic/Claude Code/bundled-skills/design-sync/lib/emit.mjs:128 Mcp skillPrompt injectionP2
high System graph security Skillspector conf 0.70 SkillSpector P2 (prompt-injection) in Anthropic/Claude Code/bundled-skills/design-sync/non-storybook/SKILL.md
<!-- @dsCard group="…" -->` for the DS pane to register it. Usually a local `lib/emit.mjs` edit dropped the header — restore it, or re-run the converter. | | `[LINK_HREF_MISSING]` | `<path>: <link hre Hidden instructions were detected in comments or invisible text. These could contain malicious di…
Anthropic/Claude Code/bundled-skills/design-sync/non-storybook/SKILL.md:93 Mcp skillPrompt injectionP2
high System graph security Skillspector conf 0.85 SkillSpector P6 (prompt-injection) in Anthropic/Claude Code/bundled-skills/design-sync/lib/css.mjs
return rules Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties. Skill: design-sync Rule: P6 Category: prompt-injection Severity: HIGH Confidence: 0.85 Remediation: Remove any instructions that reveal, prin…
Anthropic/Claude Code/bundled-skills/design-sync/lib/css.mjs:48 Mcp skillPrompt injectionP6
high System graph security Skillspector conf 0.60 SkillSpector PE3 (priv-esc) in Anthropic/Claude Code/bundled-skills/claude-api/shared/anthropic-cli.md
.env Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts. Skill: claude-api Rule: PE3 Category: priv-esc Severity: HIGH Confidence: 0.60 Remediation: Remove references to credential paths. Use environment variables or secrets managers…
Anthropic/Claude Code/bundled-skills/claude-api/shared/anthropic-cli.md:65 Mcp skillPriv escPe3
high System graph security Skillspector conf 0.70 SkillSpector PE3 (priv-esc) in Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-environments.md
Access Token Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts. Skill: claude-api Rule: PE3 Category: priv-esc Severity: HIGH Confidence: 0.70 Remediation: Remove references to credential paths. Use environment variables or secrets …
Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-environments.md:128 Mcp skillPriv escPe3
high System graph security Skillspector conf 0.70 SkillSpector PE3 (priv-esc) in Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-tools.md
Access Tokens Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts. Skill: claude-api Rule: PE3 Category: priv-esc Severity: HIGH Confidence: 0.70 Remediation: Remove references to credential paths. Use environment variables or secrets…
Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-tools.md:214 Mcp skillPriv escPe3
high System graph security Skillspector conf 0.60 SkillSpector PE3 (priv-esc) in Anthropic/Claude Code/bundled-skills/design-sync/lib/common.mjs
.env' Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts. Skill: design-sync Rule: PE3 Category: priv-esc Severity: HIGH Confidence: 0.60 Remediation: Remove references to credential paths. Use environment variables or secrets manage…
Anthropic/Claude Code/bundled-skills/design-sync/lib/common.mjs:141 Mcp skillPriv escPe3
high System graph security Skillspector conf 0.60 SkillSpector PE3 (priv-esc) in Anthropic/Claude Code/bundled-skills/design-sync/lib/css.mjs
/etc/passwd Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts. Skill: design-sync Rule: PE3 Category: priv-esc Severity: HIGH Confidence: 0.60 Remediation: Remove references to credential paths. Use environment variables or secrets …
Anthropic/Claude Code/bundled-skills/design-sync/lib/css.mjs:11 Mcp skillPriv escPe3
high System graph security Skillspector conf 0.70 SkillSpector PE3 (priv-esc) in Anthropic/Claude Code/bundled-skills/run-skill-generator/examples/electron.md
keychain Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts. Skill: run-skill-generator Rule: PE3 Category: priv-esc Severity: HIGH Confidence: 0.70 Remediation: Remove references to credential paths. Use environment variables or sec…
Anthropic/Claude Code/bundled-skills/run-skill-generator/examples/electron.md:355 Mcp skillPriv escPe3
high System graph security Skillspector conf 0.70 SkillSpector PE3 (priv-esc) in Anthropic/Claude Code/bundled-skills/run-skill-generator/SKILL.md
keychain Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts. Skill: run-skill-generator Rule: PE3 Category: priv-esc Severity: HIGH Confidence: 0.70 Remediation: Remove references to credential paths. Use environment variables or sec…
Anthropic/Claude Code/bundled-skills/run-skill-generator/SKILL.md:161 Mcp skillPriv escPe3
high System graph security Skillspector conf 0.85 SkillSpector RA1 (rogue-agent) in Anthropic/Claude Code/bundled-skills/run-skill-generator/examples/electron.md
write SKILL Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors. Skill: run-skill-generator Rule: RA1 Category: rogue-agent Severity: HIGH Confidence: 0.85 Rem…
Anthropic/Claude Code/bundled-skills/run-skill-generator/examples/electron.md:230 Mcp skillRogue agentRa1
high System graph security Skillspector conf 0.85 SkillSpector RA1 (rogue-agent) in Anthropic/Claude Code/bundled-skills/run-skill-generator/SKILL.md
Write SKILL Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors. Skill: run-skill-generator Rule: RA1 Category: rogue-agent Severity: HIGH Confidence: 0.85 Rem…
Anthropic/Claude Code/bundled-skills/run-skill-generator/SKILL.md:231 Mcp skillRogue agentRa1
high System graph security Skillspector conf 0.85 SkillSpector TM1 (tool-misuse) in Anthropic/Claude Code/bundled-skills/design-sync/lib/preview-rebuild.mjs
rm -rf's the whole bundle and must stay / Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns). Skill: design-sync Rule: TM1 Category: tool-misuse Severity: HIGH Confidence: 0.…
Anthropic/Claude Code/bundled-skills/design-sync/lib/preview-rebuild.mjs:9 Mcp skillTool misuseTm1
high System graph security Skillspector conf 0.85 SkillSpector TM1 (tool-misuse) in Anthropic/Claude Code/bundled-skills/design-sync/non-storybook/SKILL.md
rm <path>` | `--out` points at `/ Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns). Skill: design-sync Rule: TM1 Category: tool-misuse Severity: HIGH Confidence: 0.85 Reme…
Anthropic/Claude Code/bundled-skills/design-sync/non-storybook/SKILL.md:91 Mcp skillTool misuseTm1
high System graph security Skillspector conf 0.85 SkillSpector TM1 (tool-misuse) in Anthropic/Claude Code/bundled-skills/design-sync/package-build.mjs
rm -rf cwd, $HOME, /, anything in the durable / Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns). Skill: design-sync Rule: TM1 Category: tool-misuse Severity: HIGH Confiden…
Anthropic/Claude Code/bundled-skills/design-sync/package-build.mjs:190 Mcp skillTool misuseTm1
high System graph security Skillspector conf 0.85 SkillSpector TM1 (tool-misuse) in Anthropic/Claude Code/bundled-skills/run-skill-generator/examples/electron.md
rm -f /tmp/ Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns). Skill: run-skill-generator Rule: TM1 Category: tool-misuse Severity: HIGH Confidence: 0.85 Remediation: Valid…
Anthropic/Claude Code/bundled-skills/run-skill-generator/examples/electron.md:317 Mcp skillTool misuseTm1
high System graph security Skillspector conf 0.75 SkillSpector TM2 (tool-misuse) in Anthropic/Claude Code/bundled-skills/claude-api/shared/anthropic-cli.md
| sudo Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow. Skill: claude-api Rule: TM2 Category: tool-misuse Severity: HIGH Confidence: 0.75 Remediation: Limit tool chaining depth and validate the output of each tool …
Anthropic/Claude Code/bundled-skills/claude-api/shared/anthropic-cli.md:25 Mcp skillTool misuseTm2
medium System graph quality Agent instructions conf 1.00 Agent authority lacks a verifier contract: Anthropic/Claude Code/bundled-skills/design-sync/non-storybook/SKILL.md
This agent instruction grants code or shell authority but does not state the verification gate that decides promotion. The recurring safe pattern is: LLM proposes; deterministic tests/build/security checks verify; only verified code promotes.
Anthropic/Claude Code/bundled-skills/design-sync/non-storybook/SKILL.md VerificationSkill file
medium System graph quality Agent instructions conf 1.00 Agent authority lacks a verifier contract: Anthropic/Claude Code/bundled-skills/design-sync/SKILL.md
This agent instruction grants code or shell authority but does not state the verification gate that decides promotion. The recurring safe pattern is: LLM proposes; deterministic tests/build/security checks verify; only verified code promotes.
Anthropic/Claude Code/bundled-skills/design-sync/SKILL.md VerificationSkill file
medium System graph quality Agent instructions conf 1.00 Agent authority lacks a verifier contract: Anthropic/Claude Code/bundled-skills/design-sync/storybook/SKILL.md
This agent instruction grants code or shell authority but does not state the verification gate that decides promotion. The recurring safe pattern is: LLM proposes; deterministic tests/build/security checks verify; only verified code promotes.
Anthropic/Claude Code/bundled-skills/design-sync/storybook/SKILL.md VerificationSkill file
medium System graph quality Agent instructions conf 1.00 Agent authority lacks a verifier contract: Anthropic/Claude Code/bundled-skills/run-skill-generator/SKILL.md
This agent instruction grants code or shell authority but does not state the verification gate that decides promotion. The recurring safe pattern is: LLM proposes; deterministic tests/build/security checks verify; only verified code promotes.
Anthropic/Claude Code/bundled-skills/run-skill-generator/SKILL.md VerificationSkill file
medium System graph quality Agent instructions conf 1.00 Agent instructions exist but release-hardening basics are missing
AI-coder instruction files were found, but the repo is missing ci, tests. Treat this as a contract gap: the agent is guided, but the generated output is not yet guarded by the controls that make it repeatable.
Repo hardeningGenerated repo pattern
medium System graph quality Placeholder conf 1.00 Critical user flow still appears backed by mock or placeholder data
A payment/auth/admin/order/billing-style flow contains mock, fake, TODO, dummy, or placeholder markers in runtime source. In the Fable corpus this is a high-leverage completeness smell: the app can look finished while the money, identity, or tenant flow is still scaffolded.
Mock dataCritical flowGenerated repo pattern
medium System graph quality Placeholder conf 1.00 Placeholder or mock-heavy implementation detected
Found 12 placeholder/mock markers across 4 source files. This often means the repo looks complete while core flows still use generated scaffolding or fake data.
Mock dataIncompleteGenerated repo pattern
medium System graph security Skillspector conf 0.60 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/curl/examples.md
cURL / Raw HTTP Use these examples when the user needs raw HTTP requests or is working in a language without an official SDK. ## Setup ```bash export ANTHROPIC_API_KEY="your-api-key" ``` --- ## B Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. M…
Anthropic/Claude Code/bundled-skills/claude-api/curl/examples.md:1 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.60 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/curl/managed-agents.md
cURL / Raw HTTP Use these examples when the user needs raw HTTP requests or is working without an SDK. ## Setup ```bash export ANTHROPIC_API_KEY="your-api-key" # Common headers HEADERS=( -H "Con Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. M…
Anthropic/Claude Code/bundled-skills/claude-api/curl/managed-agents.md:1 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.50 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/go/managed-agents/README.md
https://api.githubcopilot.com/ Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended. Skill: claude-api Rule: E1 Category: data-exfil Severity: MEDIUM Confidence: 0.50 Remediation: Verify the destination URL is trusted and …
Anthropic/Claude Code/bundled-skills/claude-api/go/managed-agents/README.md:392 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.50 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/java/managed-agents/README.md
https://api.githubcopilot.com/ Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended. Skill: claude-api Rule: E1 Category: data-exfil Severity: MEDIUM Confidence: 0.50 Remediation: Verify the destination URL is trusted and …
Anthropic/Claude Code/bundled-skills/claude-api/java/managed-agents/README.md:303 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.50 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/php/managed-agents/README.md
https://api.anthropic.com/ Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended. Skill: claude-api Rule: E1 Category: data-exfil Severity: MEDIUM Confidence: 0.50 Remediation: Verify the destination URL is trusted and nece…
Anthropic/Claude Code/bundled-skills/claude-api/php/managed-agents/README.md:214 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.50 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/ruby/managed-agents/README.md
https://api.githubcopilot.com/ Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended. Skill: claude-api Rule: E1 Category: data-exfil Severity: MEDIUM Confidence: 0.50 Remediation: Verify the destination URL is trusted and …
Anthropic/Claude Code/bundled-skills/claude-api/ruby/managed-agents/README.md:273 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.60 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/anthropic-cli.md
curl https://api.anthropic.com/v1/messages \ -H "Authorization: Bearer $(ant auth print-credentials --access-token)" \ -H "anthropic-version: 2023-06-01" \ -H "anthropic-beta: oauth-2025-04-20" Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Ma…
Anthropic/Claude Code/bundled-skills/claude-api/shared/anthropic-cli.md:58 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.50 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-api-reference.md
https://api.githubcopilot.com/ Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended. Skill: claude-api Rule: E1 Category: data-exfil Severity: MEDIUM Confidence: 0.50 Remediation: Verify the destination URL is trusted and …
Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-api-reference.md:251 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.50 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-environments.md
https://api.githubcopilot.com/ Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended. Skill: claude-api Rule: E1 Category: data-exfil Severity: MEDIUM Confidence: 0.50 Remediation: Verify the destination URL is trusted and …
Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-environments.md:149 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.60 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-scheduled-deployments.md
curl -fsSL https://api.anthropic.com/v1/deployments \ -H "x-api-key: $ANTHROPIC_API_KEY" \ -H "anthropic-version: 2023-06-01" \ -H "anthropic-beta: managed-agents-2026-04-01" \ -H "content-typ Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. M…
Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-scheduled-deployments.md:16 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.50 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-tools.md
https://api.notion.com/ Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended. Skill: claude-api Rule: E1 Category: data-exfil Severity: MEDIUM Confidence: 0.50 Remediation: Verify the destination URL is trusted and necessa…
Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-tools.md:236 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.50 SkillSpector E1 (data-exfil) in Anthropic/Claude Code/bundled-skills/claude-api/shared/models.md
https://api.anthropic.com/ Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended. Skill: claude-api Rule: E1 Category: data-exfil Severity: MEDIUM Confidence: 0.50 Remediation: Verify the destination URL is trusted and nece…
Anthropic/Claude Code/bundled-skills/claude-api/shared/models.md:35 Mcp skillData exfilE1
medium System graph security Skillspector conf 0.85 SkillSpector EA1 (excessive-agency) in Anthropic/Claude Code/bundled-skills/claude-api/shared/agent-design.md
tool:* Skill grants unrestricted tool access without appropriate constraints. An agent with unfettered tool access can perform arbitrary actions including file modification, network requests, and code execution. Skill: claude-api Rule: EA1 Category: excessive-agency Severity: MEDIUM Confidence:…
Anthropic/Claude Code/bundled-skills/claude-api/shared/agent-design.md:26 Mcp skillExcessive agencyEa1
medium System graph security Skillspector conf 0.85 SkillSpector EA1 (excessive-agency) in Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-tools.md
tool:* Skill grants unrestricted tool access without appropriate constraints. An agent with unfettered tool access can perform arbitrary actions including file modification, network requests, and code execution. Skill: claude-api Rule: EA1 Category: excessive-agency Severity: MEDIUM Confidence:…
Anthropic/Claude Code/bundled-skills/claude-api/shared/managed-agents-tools.md:210 Mcp skillExcessive agencyEa1
medium System graph security Skillspector conf 0.85 SkillSpector EA1 (excessive-agency) in Anthropic/Claude Code/bundled-skills/claude-api/shared/tool-use-concepts.md
tools:* Skill grants unrestricted tool access without appropriate constraints. An agent with unfettered tool access can perform arbitrary actions including file modification, network requests, and code execution. Skill: claude-api Rule: EA1 Category: excessive-agency Severity: MEDIUM Confidence…
Anthropic/Claude Code/bundled-skills/claude-api/shared/tool-use-concepts.md:82 Mcp skillExcessive agencyEa1
medium System graph security Skillspector conf 0.80 SkillSpector EA2 (excessive-agency) in Anthropic/Claude Code/bundled-skills/claude-api/go/claude-api/tool-use.md
automatically run Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation. Skill: claude-api Rule: EA2 Category: excessive-agency Severity: M…
Anthropic/Claude Code/bundled-skills/claude-api/go/claude-api/tool-use.md:42 Mcp skillExcessive agencyEa2
medium System graph security Skillspector conf 0.80 SkillSpector EA2 (excessive-agency) in Anthropic/Claude Code/bundled-skills/claude-api/python/claude-api/tool-use.md
automatically run Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation. Skill: claude-api Rule: EA2 Category: excessive-agency Severity: M…
Anthropic/Claude Code/bundled-skills/claude-api/python/claude-api/tool-use.md:28 Mcp skillExcessive agencyEa2
medium System graph security Skillspector conf 0.75 SkillSpector EA2 (excessive-agency) in Anthropic/Claude Code/bundled-skills/claude-api/shared/model-migration.md
without asking Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation. Skill: claude-api Rule: EA2 Category: excessive-agency Severity: MEDI…
Anthropic/Claude Code/bundled-skills/claude-api/shared/model-migration.md:1228 Mcp skillExcessive agencyEa2
medium System graph security Skillspector conf 0.80 SkillSpector EA2 (excessive-agency) in Anthropic/Claude Code/bundled-skills/claude-api/SKILL.md
don't ask the user Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation. Skill: claude-api Rule: EA2 Category: excessive-agency Severity: …
Anthropic/Claude Code/bundled-skills/claude-api/SKILL.md:208 Mcp skillExcessive agencyEa2
medium System graph security Skillspector conf 0.75 SkillSpector EA2 (excessive-agency) in Anthropic/Claude Code/bundled-skills/design-sync/lib/sync-hashes.mjs
no verification Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation. Skill: design-sync Rule: EA2 Category: excessive-agency Severity: ME…
Anthropic/Claude Code/bundled-skills/design-sync/lib/sync-hashes.mjs:100 Mcp skillExcessive agencyEa2
medium System graph security Skillspector conf 0.75 SkillSpector EA2 (excessive-agency) in Anthropic/Claude Code/bundled-skills/design-sync/non-storybook/SKILL.md
without asking Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation. Skill: design-sync Rule: EA2 Category: excessive-agency Severity: MED…
Anthropic/Claude Code/bundled-skills/design-sync/non-storybook/SKILL.md:20 Mcp skillExcessive agencyEa2
medium System graph security Skillspector conf 0.80 SkillSpector EA4 (excessive-agency) in Anthropic/Claude Code/bundled-skills/claude-api/python/claude-api/README.md
max_retries=0 Skill allows unbounded resource consumption (API calls, storage, compute). Without rate limits or quotas, a compromised or misbehaving agent can cause denial-of-service or cost overruns. Skill: claude-api Rule: EA4 Category: excessive-agency Severity: MEDIUM Confidence: 0.80 Reme…
Anthropic/Claude Code/bundled-skills/claude-api/python/claude-api/README.md:57 Mcp skillExcessive agencyEa4
medium System graph security Skillspector conf 0.80 SkillSpector MP2 (memory-poisoning) in Anthropic/Claude Code/bundled-skills/claude-api/shared/agent-design.md
exceed the context window Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries. Skill: claude-api Rule: MP2 Category: memory-poisoning Severity: MEDIUM Confidence:…
Anthropic/Claude Code/bundled-skills/claude-api/shared/agent-design.md:80 Mcp skillMemory poisoningMp2
medium System graph security Skillspector conf 0.80 SkillSpector MP2 (memory-poisoning) in Anthropic/Claude Code/bundled-skills/claude-api/SKILL.md
exceed context window Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries. Skill: claude-api Rule: MP2 Category: memory-poisoning Severity: MEDIUM Confidence: 0.8…
Anthropic/Claude Code/bundled-skills/claude-api/SKILL.md:468 Mcp skillMemory poisoningMp2
medium System graph security Skillspector conf 0.85 SkillSpector MP2 (memory-poisoning) in Anthropic/Claude Code/bundled-skills/run-skill-generator/examples/playwright.md
fill input Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries. Skill: run-skill-generator Rule: MP2 Category: memory-poisoning Severity: MEDIUM Confidence: 0.85 …
Anthropic/Claude Code/bundled-skills/run-skill-generator/examples/playwright.md:35 Mcp skillMemory poisoningMp2
medium System graph security Skillspector conf 0.70 SkillSpector PE2 (priv-esc) in Anthropic/Claude Code/bundled-skills/claude-api/shared/anthropic-cli.md
sudo Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified. Skill: claude-api Rule: PE2 Category: priv-esc Severity: MEDIUM Confidence: 0.70 Remediation: Avoid sudo/root unless strictly required. Prefer least-privilege patterns. If elevation is needed,…
Anthropic/Claude Code/bundled-skills/claude-api/shared/anthropic-cli.md:25 Mcp skillPriv escPe2
medium System graph security Skillspector conf 0.60 SkillSpector RA2 (rogue-agent) in Anthropic/Claude Code/bundled-skills/claude-api/SKILL.md
write `claude-fable-5` code, include the server-side `fallbacks` parameter by default** (`betas: ["server-side-fallback-2026-06-01"]` + `fallbacks: [{"model": "claude-opus-4-8"}]`; Claude API and Clau Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or sta…
Anthropic/Claude Code/bundled-skills/claude-api/SKILL.md:193 Mcp skillRogue agentRa2
medium System graph security Skillspector conf 0.75 SkillSpector RA2 (rogue-agent) in Anthropic/Claude Code/bundled-skills/design-sync/lib/source-kit.mjs
defaults write Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction. Skill: design-sync Rule: RA2 Category: rogue-agent Severity: MEDIUM Confidence: 0.7…
Anthropic/Claude Code/bundled-skills/design-sync/lib/source-kit.mjs:9 Mcp skillRogue agentRa2
medium System graph security Skillspector conf 0.75 SkillSpector RA2 (rogue-agent) in Anthropic/Claude Code/bundled-skills/design-sync/SKILL.md
defaults write Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction. Skill: design-sync Rule: RA2 Category: rogue-agent Severity: MEDIUM Confidence: 0.7…
Anthropic/Claude Code/bundled-skills/design-sync/SKILL.md:62 Mcp skillRogue agentRa2
medium System graph quality Tests conf 1.00 Very low test-to-source ratio
0 test file(s) for 28 source file(s) (ratio 0.00). Consider adding integration or unit tests for critical paths.
Coverage
low System graph quality Production readiness conf 1.00 Composite production-readiness gap
Multiple low-cost hardening controls are missing together: ci, tests. Opus verification showed these co-occurring gaps are a better readiness signal than reading each flag in isolation.
Repo hardeningGenerated repo pattern
low System graph quality Debug conf 1.00 Debug logging residue appears in source files
Found 14 console/debugger/print-style debug statements in non-test source. This is a common fast-generation residue before production cleanup.
CleanupRepo hardeningGenerated repo pattern
low System graph software Dead code conf 1.00 Possibly dead Python function: lin2s
No callers detected by AST scan in this repo. Could be exported for external callers or a framework handler.
Anthropic/Claude Code/bundled-skills/dataviz/scripts/validate_palette.py:87
low System graph security Skillspector conf 0.70 SkillSpector EA3 (excessive-agency) in Anthropic/Claude Code/bundled-skills/claude-api/LICENSE.txt
not limited to Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface. Skill: claude-api Rule: EA3 Category: excessive-agency Severity: LOW Confidence: 0.70 Remed…
Anthropic/Claude Code/bundled-skills/claude-api/LICENSE.txt:28 Mcp skillExcessive agencyEa3
low System graph quality Complexity conf 1.00 Very large file: Anthropic/Claude Code/bundled-skills/design-sync/package-build.mjs (1005 lines)
Files with >800 lines often hide complexity hotspots and discourage tests.
low System graph quality Complexity conf 1.00 Very large file: Anthropic/Claude Code/bundled-skills/design-sync/package-validate.mjs (804 lines)
Files with >800 lines often hide complexity hotspots and discourage tests.
For AI agents: Voting guide (TP/FP) MCP manifest Stdio wrapper SARIF Integrate Findings queue Vote TP/FP on findings to calibrate the engine.
For AI agents + API integrations
Email me when this repo regresses
Free. We re-scan periodically; new criticals → your inbox. No signup required for the scan itself.
API access

This page is publicly accessible at: https://repobility.com/scan/f2c21032-c9c3-443a-8cc0-e029eaeba567/

To check status programmatically (no auth required): 

curl -s https://repobility.com/api/v1/public/scan/f2c21032-c9c3-443a-8cc0-e029eaeba567/

Important — please don't re-submit the same URL repeatedly. The submission endpoint is idempotent: re-submitting the same git URL returns this same scan_token, not a new one. To re-scan this repo, sign up free and use the dashboard.