Please do not report security vulnerabilities through public GitHub issues.

Report vulnerabilities using one of these channels:

1. GitHub Private Advisory (preferred): Create a private advisory
2. Email: sairam.ugge@gmail.com — subject line [SECURITY] Inkforge vulnerability

Please include:

• Type of vulnerability (e.g. credential exposure, prompt injection, SSRF)
• Affected file paths and line numbers
• Steps to reproduce
• Potential impact

Response SLA: Acknowledgment within 48 hours · Resolution timeline within 7 days.

Inkforge handles several external API credentials (AWS Bedrock, Anthropic, Dev.to, Hashnode).

• All credentials are loaded exclusively from environment variables — never hardcoded
• .env is gitignored and must never be committed
• See .env.example for the complete list of required variables
• Base64 credential handling in packages/core/src/llm/index.ts uses round-trip equality verification to prevent accidental decoding of non-base64 values
• The bedrockCreds() function uses BEDROCK_REGION instead of AWS_REGION to avoid Vercel/Lambda platform mangling of reserved variable names

• User input is forwarded to AWS Bedrock or Anthropic API — review your provider's data handling and privacy policies before processing sensitive content
• Generated content is written to the local filesystem only — no external storage by default
• No prompt injection prevention is implemented beyond standard LLM system prompt guardrails
• Do not pass secrets, PII, or confidential data as input to the generation pipeline

• Dev.to and Hashnode API keys are used solely for publishing — stored in .env
• Published content sets canonical_url pointing back to your portfolio — cross-posting is SEO-safe
• The /api/config endpoint in @inkforge/web returns only boolean flags — never exposes actual key values

• All dependencies are pinned in pnpm-lock.yaml
• Run pnpm audit regularly to check for known vulnerabilities
• A moderate severity vulnerability exists in postcss (via next@16.2.9) — no upstream fix available at this version; it affects CSS processing only and is not exploitable in Inkforge's context

• CI workflows do not have access to production secrets
• Dependency audit runs on every push and PR
• Secret scanning runs on every push via the built-in workflow

• Vulnerabilities in upstream LLM providers (AWS Bedrock, Anthropic)
• AI-generated article content
• Third-party publishing platforms (Dev.to, Hashnode, Medium, Substack)
• Issues in node_modules that are not reachable through Inkforge's code paths