Skip to content

For the complete documentation index, see llms.txt.

Frequently asked questions

Answers to common questions about Promptless for Agent Instructions. Each answer links to the page that covers the topic in full.

Your host uploads verbatim native trace bytes and batch metadata only to the worker you run in your own infrastructure. Hosted Promptless receives a limited metadata projection (identifiers, counts, and status), never trace content. See Trust and data model.

Promptless compiles plugins for four targets (Claude, Codex, Cursor, and Gemini) and ingests session traces from three sources (Claude Code, Claude Desktop, and Codex). See Supported agents.

Human review. A remediation agent opens a pull request against your Instruction Hub, and a person reviews and merges it. There is no automated evaluation gate today. See Remediate findings.

The pig toolchain is the public Python package promptless-instruction-hub, in the Promptless/instruction-hub-toolchain repository. It runs in your continuous-integration pipeline, not as a service. See Set up your Instruction Hub.

A Postgres database and an S3 bucket you own, plus a plih_ install token. Before the worker can produce findings, you also configure a model provider through the INSTRUCTION_HUB_ANALYSIS_MODEL_* variables, since the Friction Analyzer sends its analysis input there. You deploy the worker into your own infrastructure with Docker and Helm. See Deploy the analyzer worker and Configuration reference.

Does hosted Promptless see my GitHub credentials?

Section titled “Does hosted Promptless see my GitHub credentials?”

No. The worker never receives GitHub credentials. Findings are projected to GitHub issues by hosted Promptless, and remediation runs with a repository-scoped token in an isolated agent. See Trust and data model.

If you do not see your question answered here, reach our support team at help@gopromptless.ai. During onboarding, we also set up a shared Slack Connect channel between your team and ours for direct support.