This drops a sanitized agent session log into your GitHub PR or issue body so reviewers can see what the bot actually did. It scans local JSONL logs from Claude, Codex, or OpenClaw sessions, strips out secrets and system prompts, then gives you an HTML preview before appending a collapsible transcript section. The contract is strict: never hits the network, fails closed on auth leaks, and auto-trims unrelated turns by comparing against the PR title and changed files. It's best-effort provenance, not a guarantee. If you're running agent workflows that cut PRs and want reviewers to trust the work without reading every line of diff, this makes the history visible without manual copypaste.
npx -y skills add openclaw/agent-skills --skill agent-transcript --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
github/awesome-copilot
microsoft/win-dev-skills
github/awesome-copilot
github/awesome-copilot