agent-lessons

Preserve and retrieve reusable context in a Markdown knowledge base.

Establish or refresh the workspace resource map and personal workflow in a managed section in root AGENTS.md.

Help a human learn or understand a topic, technical system, code change, error, agent output, or source material.

Implements, debugs, fixes, refactors, and reviews code.

Use during execution when progress is blocked, or likely to stall, on an action or information only the human can provide—for example access approval…

Maintains settled project understanding in the documentation surface explicitly configured in applicable AGENTS.md, and stable agent behavior in explicitly…

Manage and execute Python through reusable, independent uv purpose projects under the Python Environment Root explicitly configured in AGENTS.md.

Maintains shared human-agent continuation briefs in the task surface explicitly configured in applicable AGENTS.md (including tasks/backlog/ for deferred…

Improve human readability for readability-focused requests: rewrite a hard-to-read response or document, review its readability, or make a new answer or…

Evaluate, strengthen, and communicate a position through sound reasoning, evidence, alternatives, and counterarguments.

Perform a decision-focused ML architecture investigation with evidence, alternatives, trade-offs, risks, and a leading recommendation.

Extract text from a specified PDF or image into local temporary Markdown and a quality report only when the human explicitly requests extraction.