mangi-11/eta/self-improving-agent

Self Improving Agent

This built-in skill is fixed-injected for agent runs.

Use it to maintain a lightweight learning loop without interrupting the user's main task.

The runtime may auto-read this skill after a tool failure and auto-record the failure into data/ERRORS.md.

When To Record

Record after the immediate task is safe or complete when any of these happens:

1. a non-trivial command, tool, or device action fails
2. the user corrects your understanding, path, rule, or project assumption
3. you discover an outdated runtime/project convention
4. you find a reusable workaround or best practice that will likely save future retries
5. the same mistake repeats in the same task or across tasks

Do not record ordinary chat, tiny one-off slips, or anything the user asked not to save.

Default Storage

• skill-local learnings: self-improving-agent/data/
• error log: self-improving-agent/data/ERRORS.md (auto-managed by the failure hook)

Logging Workflow

1. Finish or stabilize the current user-facing step first.
2. The runtime automatically logs structured errors after tool failures — you do not need to manually write to ERRORS.md.
3. Use skill scope by default.
4. Use learning for corrected knowledge or best practices.
5. Use error for concrete failures with stderr, HTTP errors, stack traces, or invalid assumptions.
6. Use feature for recurring capability gaps the user actually wants.

Memory Promotion

Promote a lesson into a stable rule only when it is short, reusable, and broadly applicable.

Good candidates:

• a rule like "遇到 X 先检查 Y"
• a stable workspace convention
• a long-term user preference the user explicitly wants remembered

Prefer this order:

1. errors are auto-logged into the skill data by the failure hook
2. review data/ERRORS.md when similar failures recur
3. distill stable rules into your system prompt or user-facing guidance

Output Discipline

• keep summaries short and specific
• include the concrete command/tool/context that failed
• include the corrected rule, not only the symptom
• avoid logging secrets, tokens, and personal data