0.1

First public release of Signal Flare as a Claude skill.

Status: experimental. The framework is sound; platform-specific details vary in maturity. See STATUS.md for per-platform confidence levels.

Install

Download signal-flare.skill below and install it in Claude. Once installed, start a conversation by describing your situation — for example: "My Discord account just got hijacked and Discord support keeps closing my tickets. Can you help me get it back?" — and Claude will walk you through the framework.

What's in v0.1

• Conversational skill flow — Claude triages your scenario, walks you through the relevant escalation stages, constructs each escalation document tailored to your specific case, stress-tests them against adversarial perspectives, and prepares you for an attorney conversation if it gets that far.
• 5-stage escalation framework (FRAMEWORK.md) — Stage 0 stop-the-bleeding, Stage 1 internal escalation, Stage 2 asset preservation, Stage 3 regulatory pressure (state AG / FTC / congressional / BBB), Stage 4 attorney demand letter, Stage 5 litigation.
• 12 platform addendums (platforms/) — Discord, Meta, and PayPal verified; the other 9 (Google, X, TikTok, Twitch, LinkedIn, Snapchat, Reddit, Roblox, Steam) are initial drafts that need independent verification before relying on platform-specific contact details.
• 8 construction briefs (templates/) — not fill-in-the-blank templates, but briefs that tell Claude what each escalation document needs to do, what always goes in, what varies by scenario, and what to avoid. The construction-brief approach produces documents that vary case-by-case rather than reading as form-letter activism.
• Conversational ethical-floor gate — the skill determines through real conversation whether the user is the legitimate owner, can prove ownership, and didn't commit the violation they're appealing. If the case is the wrong fit, the skill declines and points elsewhere.
• Hard constraints against hallucinated legal advice — Claude is instructed never to name specific legal theories, statutes, case law, or dollar values. That's the attorney's job.
• Outline-first protocol on high-stakes documents — for state AG complaints, attorney prep, legal notices, congressional letters, and FTC complaints, Claude produces a structured outline and confirms it with the user before writing prose. This catches hallucinated facts before they reach a regulator.
• Standalone fallback guide (using-ai-to-build-your-case.md) — for users who don't have Claude or who want to use ChatGPT/Gemini instead. Same prompts the skill encapsulates, formatted for paste-into-chat use.
• Maintenance discipline (MAINTENANCE.md) — quarterly review cadence, last-verified markers, stale-warning thresholds at 180 / 365 / 540 days, sunset clause if maintenance lapses.
• Automated link checking — weekly GitHub Actions workflow opens issues for broken URLs and fails PRs that introduce new ones.

Read this first

This skill only helps if all three are true:

1. You are the legitimate owner of the account
2. You can prove it with documentation
3. The platform's standard support has failed you

It will not help with: recovering an account someone else owned (former partner, deceased relative, former employer, account purchased from a reseller), reversing a ban for a Terms of Service violation you actually committed, or "social engineering" past identity verification.

The framework's effectiveness depends on every claim you file being true. Filing knowingly false statements with state Attorneys General, the FTC, or the courts is a crime. The skill is built around that principle and refuses to help when it would compromise it.

Honest expectations

This framework maximizes the chance of resolution. It does not guarantee one. Platforms can refuse. Regulators can decline to forward complaints. Attorneys can decline to take cases. Some cases — particularly ones with weak documentation, or where the platform has substantive (not pretextual) reasons for the action — are unsalvageable. Most resolutions take 2–6 weeks and arrive quietly.

Known limitations

• 9 of 12 platform addendums are initial drafts. Verify platform-specific details (URLs, email addresses, support procedures) against current sources before sending anything to those addresses. Discord, Meta, and PayPal are the most thoroughly researched.
• Stage 3 regulatory pathways are US-specific. State AGs, FTC, congressional offices apply only to US residents. Non-US users need substitute mechanisms (their country's data protection authority, consumer protection regulator, or telecommunications ombudsman) — Stages 0, 1, 2, 4, and 5 generalize fine, but Stage 3 needs adaptation. Contributions documenting non-US equivalents are welcome.
• Real anonymized precedents and regional adaptations are placeholders awaiting community contributions. The framework is documented; the case stories and non-US guides are not yet populated.

License

Apache 2.0. Use it, fork it, adapt it, ship it elsewhere — all permitted with attribution.