self-improving agent

- Added detailed instructions and format standards for logging learnings, errors, and feature requests to markdown files in a `.learnings/` directory. - Introduced a robust initialization process to automatically create missing `.learnings` log files without overwriting existing data. - Provided clear tables and examples on when and how to log events, escalate learnings, and promote common patterns to broader workspace files. - Added setup guidance for use in OpenClaw and generic agent environments, with installation, directory structures, and workflow integration steps. - Outlined security best practices—never log secrets or raw output by default; use concise, redacted summaries instead. - Offered sample logging templates and metadata fields to standardize entries for continuous self-improvement.

- Initial release for OpenClaw: rewritten from the original self-improvement agent to fit OpenClaw workspace and workflows. - Adds automatic logging of learnings, errors, and feature requests to structured markdown files in `.learnings/`. - Provides step-by-step initialization and logging instructions, preventing overwriting or logging of secrets by default. - Includes quick reference table to map common situations to correct log targets. - Documents methods to promote important learnings to wider project/workspace files (`AGENTS.md`, `SOUL.md`, `TOOLS.md`, etc.). - Offers OpenClaw-specific integration (hooks, session sharing tools), with safe practices for cross-session knowledge transfer.

re-upload

re-upload no changes

no changes added re-upload

No code or documentation changes detected in this release. - Version bump only; no functional changes. - All files remain unchanged from the previous version.

Fixed duplicate SELF_IMPROVEMENT_REMINDER.md bootstrap injection in the OpenClaw hook. Hardened hook behavior to skip sub-agent sessions and avoid overwriting non-hook files at the same path. Improved reminder safety guidance to avoid logging secrets, tokens, env vars, or raw transcripts. Added safer first-use initialization guidance for .learnings/ files. Moved bundled log templates from .learnings/ into assets/ to avoid committing local learning logs by accident. Strengthened privacy and documentation guidance across the skill. Cleaned up repeated end-of-file guidance in SKILL.md to keep the skill definition leaner and less repetitive.

Fixed duplicate SELF_IMPROVEMENT_REMINDER.md bootstrap injection in the OpenClaw hook. Hardened hook behavior to skip sub-agent sessions and avoid overwriting non-hook files at the same path. Improved reminder safety guidance to avoid logging secrets, tokens, env vars, or raw transcripts. Added safer first-use initialization guidance for .learnings/ files. Moved bundled log templates from .learnings/ into assets/ to avoid committing local learning logs by accident. Strengthened privacy and documentation guidance across the skill. Cleaned up repeated end-of-file guidance in SKILL.md to keep the skill definition leaner and less repetitive.

re-upload no changes

re-upload no changes

re-upload no changes

Self-improving-agent v3.0.10 introduces first-use initialization and improved privacy protections. - Added automatic setup guidance and shell commands to ensure `.learnings/` files exist before logging. - Removed default logging of full transcripts, secrets, or sensitive data; added explicit privacy and redaction warnings. - Migrated learning log templates and setup steps into SKILL.md, eliminating separate `.learnings/*.md` assets. - Clarified guidance for trusted use of inter-session learning sharing. - No breaking changes to main usage or logging formats.

removed empty metadata nothing else re-upload

No code or documentation changes detected in this release. - Version update with no modifications to files or SKILL.md. - No user-facing changes or new features.

no change re-upload

no changes re-upload after clawhub update

no changes re-upload after clawhub update

no changes re-uploaded after vanishing from clawhub

**New: Comprehensive guidelines for continuous self-improvement logging and promotion across OpenClaw and generic agent setups.** - Detailed instructions for logging errors, learnings, and feature requests to markdown files with purpose-built templates. - Quick-reference tables for when and where to log different types of events (errors, corrections, feature requests, best practices). - Clear process for promoting broadly useful learnings to project memory files (CLAUDE.md, AGENTS.md, SOUL.md, TOOLS.md, etc.). - Step-by-step setup guidance for using the skill with OpenClaw (recommended) or other agent platforms. - Fully documented standardized markdown formats for each log type, including metadata and resolution tracking. - Includes workflow tips for session linking, daily memory integration, and optional hook reminders.

Version 3.0.1 – Expanded documentation, clarified workflow, and OpenClaw integration - Major overhaul of SKILL.md with detailed instructions for logging errors, learnings, and feature requests. - Added comprehensive quick reference tables for when and how to log issues and promote learnings. - Included complete setup steps for OpenClaw and generic (non-OpenClaw) agents, with directory structure and file descriptions. - Standardized formats for recording learnings, errors, and feature requests, including metadata and resolution process. - Documented promotion workflow to elevate important learnings to project memory files. - Clarified inter-session communication tools within OpenClaw for sharing and reviewing learnings.