publish-evidence

To publish cryptographically signed evidence, this Skill packages your final Claude Code prompts, responses, and MCP tool calls into a timestamped payload with provenance metadata, then uploads it to the civicaitools.org registry via the /api/evidence endpoint.

Data provenance tracks the origin and history of your civic-data analysis. It is needed to provide end-to-end integrity guarantees, ensuring that government, journalism, and research outputs remain auditable and verifiable against tampering.

When publishing verifiable evidence with large conversation transcripts, the Skill optionally uploads large fields to blob storage while preserving inline content and cryptographic hashes for integrity validation within the main evidence package.

Yes, you can include full multi-turn conversation transcripts in the published evidence package. The Skill captures the complete transcript when requested, alongside tool-call provenance and identity metadata, to ensure complete verifiable context.

Yes, you need authentication credentials to publish evidence to the civicaitools.org registry. The Skill validates your credentials before enforcing the payload schema and submitting the signed package, with support for authentication fallback mechanisms.

A key limitation of using blob storage for large evidence payloads is that large fields are offloaded externally, meaning verification requires retrieving the blob separately while relying on preserved inline hashes for integrity rather than direct inline content validation.