每日备份 2026-06-10 12:57:13

2026-06-10 12:57:14 +08:00
1 parent fda2ccb8c2
commit 3d2e638c6a
@@ -0,0 +1,66 @@
+# Domain Config for: [skill-name]
+
+Fill this template when extracting domain from a target skill. Use the extracted values to parameterize LEARNINGS.md, ERRORS.md, FEATURE_REQUESTS.md, and EVOLUTION.md.
+
+## Capability & Scenario Analysis (before extraction)
+
+**Capabilities** (what the skill does):
+- Primary outputs:
+- Secondary/edge capabilities:
+- Dependencies:
+
+**Scenarios** (when and how it is used):
+- User personas:
+- Typical tasks:
+- Input/output patterns:
+
+**Evolution directions** (what can improve):
+- User feedback patterns:
+- Failure modes:
+- Recurring corrections → domain rules:
+
+---
+
+## Domain
+
+- **Name**: (human-readable domain name, e.g., 洗稿能手, 电脑加速)
+- **Use cases**: (comma-separated scenarios, e.g., 科普、汇报、直播)
+- **Areas**: (domain-specific areas for filtering, e.g., 文案 | 口播 | 短视频脚本)
+- **Language**: (target language for generated files: zh | en)
+
+## Learning Triggers (user feedback phrases)
+
+Phrases that indicate user correction or style mismatch → log to LEARNINGS.md:
+
+- "phrase1"
+- "phrase2"
+- "phrase3"
+
+## Error Triggers (failure modes)
+
+Situations that indicate operation failure → log to ERRORS.md:
+
+- failure1
+- failure2
+- failure3
+
+## Promotion Targets
+
+Files to promote learnings to when they prove broadly applicable:
+
+- **专属进化规则**: [skill-name]-专属进化规则.md
+- **最佳实践 / 安全规范**: [skill-name]-最佳实践.md (or [skill-name]-安全规范.md)
+
+## Domain Categories (optional)
+
+If the skill needs categories beyond correction | insight | knowledge_gap | best_practice:
+
+- category1
+- category2
+
+## Notes
+
+- Use kebab-case for skill-name in file paths
+- Areas should partition the skill's output/workflow scope
+- Add 3+ learning triggers and 2+ error triggers for effective capture
+- Storage: same as self-improving-agent — `.learnings/` directly (LEARNINGS.md, ERRORS.md, FEATURE_REQUESTS.md)
@@ -0,0 +1,50 @@
+# Errors Log
+
+{DOMAIN_DESCRIPTION}
+
+**Areas**: {DOMAIN_AREAS}
+**Statuses**: pending | in_progress | resolved | wont_fix
+
+## Error Entry Format
+
+Append entries in this format:
+
+```markdown
+## [ERR-YYYYMMDD-XXX] operation_or_error_type
+
+**Logged**: ISO-8601 timestamp
+**Priority**: high
+**Status**: pending
+**Area**: {DOMAIN_AREAS_EXAMPLE}
+
+### Summary
+Brief description of what failed
+
+### Error
+```
+Actual error message or output
+```
+
+### Context
+- Operation attempted
+- Input or parameters used
+- Environment details if relevant
+
+### Suggested Fix
+If identifiable, what might resolve this
+
+### Metadata
+- Reproducible: yes | no | unknown
+- Related Files: path/to/file.ext
+- See Also: ERR-YYYYMMDD-XXX (if recurring)
+
+---
+```
+
+## Placeholders to Replace
+
+- `{DOMAIN_DESCRIPTION}`: e.g., "Operation failures, exceptions, and unexpected behaviors for [skill-name]."
+- `{DOMAIN_AREAS}`: e.g., "文案 | 口播 | 短视频脚本" or "系统优化 | 卡顿 | 报错"
+- `{DOMAIN_AREAS_EXAMPLE}`: One example area from the list
+
+---
@@ -0,0 +1,100 @@
+# [Skill Name] Evolution Rules
+
+Domain-specific self-improvement rules for [skill-name]. Log learnings and errors to `.learnings/`; promote proven patterns to the targets below. **Evolution is driven by user feedback.**
+
+## Quick Reference
+
+| Situation | Action |
+|-----------|--------|
+| {LEARNING_TRIGGER_1} | Log to `.learnings/LEARNINGS.md` with category `correction` or `style_mismatch` |
+| {LEARNING_TRIGGER_2} | Log to `.learnings/LEARNINGS.md` |
+| {ERROR_TRIGGER_1} | Log to `.learnings/ERRORS.md` |
+| {ERROR_TRIGGER_2} | Log to `.learnings/ERRORS.md` |
+| User wants missing capability | Log to `.learnings/FEATURE_REQUESTS.md` |
+| Pattern proven across 3+ instances | Promote to {PROMOTION_TARGET_1} |
+| Safety/correctness rule | Promote to {PROMOTION_TARGET_2} |
+
+## Review → Apply → Report Loop
+
+**Before task**: Load relevant entries from `.learnings/LEARNINGS.md` (and ERRORS.md if applicable). Filter by area, tags, or keywords.
+
+**During task**: Apply learnings when relevant. Optionally annotate: "本次参考了 [LRN-xxx]: ..." (or equivalent in target language).
+
+**After task**: Tell user which learnings were used, what evolution result, what improvement. Decide per-use mention vs end-of-task summary based on context.
+
+## Detection Triggers
+
+### Learning Triggers (→ LEARNINGS.md)
+
+- {LEARNING_TRIGGER_1}
+- {LEARNING_TRIGGER_2}
+- {LEARNING_TRIGGER_3}
+- User provides correction or preference
+- Better approach discovered for recurring task
+
+### Error Triggers (→ ERRORS.md)
+
+- {ERROR_TRIGGER_1}
+- {ERROR_TRIGGER_2}
+- {ERROR_TRIGGER_3}
+- Operation fails unexpectedly
+- Unexpected output or behavior
+
+## Domain Activation Conditions (for hooks)
+
+When to remind the agent to check learnings:
+
+- {ACTIVATION_CONDITION_1}
+- {ACTIVATION_CONDITION_2}
+
+## Promotion Decision Tree
+
+```
+Is the learning skill-specific?
+├── Yes → Keep in .learnings/
+└── No → Is it style/behavior-related?
+    ├── Yes → Promote to {PROMOTION_TARGET_1}
+    └── No → Is it safety/correctness-related?
+        ├── Yes → Promote to {PROMOTION_TARGET_2}
+        └── No → Promote to {PROMOTION_TARGET_1}
+```
+
+## Experience Invalidation & Update
+
+When user corrects again after a learning was applied:
+
+- Add `Contradicted-By: LRN-YYYYMMDD-XXX` to the original entry
+- Mark `Status: superseded` or `Last-Valid: YYYY-MM-DD` if the learning is no longer valid
+- Increment `Recurrence-Count` if the pattern recurs but the fix is different
+
+## Area Tags
+
+| Area | Scope |
+|------|-------|
+| {AREA_1} | {AREA_1_SCOPE} |
+| {AREA_2} | {AREA_2_SCOPE} |
+| {AREA_3} | {AREA_3_SCOPE} |
+
+## Priority Guidelines
+
+| Priority | When to Use |
+|----------|-------------|
+| `critical` | Blocks core functionality, safety risk |
+| `high` | Significant impact, affects common workflows |
+| `medium` | Moderate impact, workaround exists |
+| `low` | Minor inconvenience, edge case |
+
+## Promotion Targets
+
+- **{PROMOTION_TARGET_1}**: Evolution rules, style preferences, best practices
+- **{PROMOTION_TARGET_2}**: Safety guidelines, correctness rules (if applicable)
+
+## Placeholders to Replace
+
+- `[skill-name]`, `[Skill Name]`: Domain name
+- `{LEARNING_TRIGGER_1}`, `{LEARNING_TRIGGER_2}`, `{LEARNING_TRIGGER_3}`: User feedback phrases from domain config
+- `{ERROR_TRIGGER_1}`, `{ERROR_TRIGGER_2}`, `{ERROR_TRIGGER_3}`: Failure modes from domain config
+- `{ACTIVATION_CONDITION_1}`, `{ACTIVATION_CONDITION_2}`: Domain-specific activation (e.g., "User says 改得不通顺", "Optimization reported无效")
+- `{PROMOTION_TARGET_1}`, `{PROMOTION_TARGET_2}`: e.g., `[skill-name]-专属进化规则.md`, `[skill-name]-最佳实践.md`
+- `{AREA_1}`, `{AREA_2}`, `{AREA_3}`: Domain areas
+- `{AREA_1_SCOPE}`, etc.: Brief scope description for each area
@@ -0,0 +1,45 @@
+# Feature Requests
+
+{DOMAIN_DESCRIPTION}
+
+**Areas**: {DOMAIN_AREAS}
+**Statuses**: pending | in_progress | resolved | wont_fix
+
+## Feature Request Entry Format
+
+Append entries in this format:
+
+```markdown
+## [FEAT-YYYYMMDD-XXX] capability_name
+
+**Logged**: ISO-8601 timestamp
+**Priority**: medium
+**Status**: pending
+**Area**: {DOMAIN_AREAS_EXAMPLE}
+
+### Requested Capability
+What the user wanted to do
+
+### User Context
+Why they needed it, what problem they're solving
+
+### Complexity Estimate
+simple | medium | complex
+
+### Suggested Implementation
+How this could be built, what it might extend
+
+### Metadata
+- Frequency: first_time | recurring
+- Related Features: existing_feature_name
+
+---
+```
+
+## Placeholders to Replace
+
+- `{DOMAIN_DESCRIPTION}`: e.g., "Capabilities requested by user that don't currently exist for [skill-name]."
+- `{DOMAIN_AREAS}`: e.g., "文案 | 口播 | 短视频脚本" or "系统优化 | 卡顿 | 报错"
+- `{DOMAIN_AREAS_EXAMPLE}`: One example area from the list
+
+---
@@ -0,0 +1,62 @@
+# Learnings
+
+{DOMAIN_DESCRIPTION}
+
+**Categories**: {DOMAIN_CATEGORIES}
+**Areas**: {DOMAIN_AREAS}
+**Statuses**: pending | in_progress | resolved | wont_fix | promoted | promoted_to_skill | superseded
+
+## Status Definitions
+
+| Status | Meaning |
+|--------|---------|
+| `pending` | Not yet addressed |
+| `in_progress` | Actively being worked on |
+| `resolved` | Issue fixed or knowledge integrated |
+| `wont_fix` | Decided not to address (reason in Resolution) |
+| `promoted` | Elevated to {PROMOTION_TARGETS} |
+| `promoted_to_skill` | Extracted as a reusable skill |
+| `superseded` | User corrected again; this learning is no longer valid |
+
+## Learning Entry Format
+
+Append entries in this format:
+
+```markdown
+## [LRN-YYYYMMDD-XXX] category
+
+**Logged**: ISO-8601 timestamp
+**Priority**: low | medium | high | critical
+**Status**: pending
+**Area**: {DOMAIN_AREAS_EXAMPLE}
+
+### Summary
+One-line description of what was learned
+
+### Details
+Full context: what happened, what was wrong, what's correct
+
+### Suggested Action
+Specific fix or improvement to make
+
+### Metadata
+- Source: conversation | error | user_feedback
+- Related Files: path/to/file.ext
+- Tags: tag1, tag2
+- See Also: LRN-YYYYMMDD-XXX (if related to existing entry)
+- Recurrence-Count: 1 (optional, bump when pattern recurs)
+- Last-Valid: YYYY-MM-DD (optional, when user contradicted)
+- Contradicted-By: LRN-YYYYMMDD-XXX (optional, when superseded)
+
+---
+```
+
+## Placeholders to Replace
+
+- `{DOMAIN_DESCRIPTION}`: e.g., "Corrections, insights, and style preferences captured for [skill-name]."
+- `{DOMAIN_CATEGORIES}`: e.g., "correction | insight | knowledge_gap | best_practice | style_mismatch | scene_adaptation"
+- `{DOMAIN_AREAS}`: e.g., "文案 | 口播 | 短视频脚本" or "系统优化 | 卡顿 | 报错"
+- `{DOMAIN_AREAS_EXAMPLE}`: One example area from the list
+- `{PROMOTION_TARGETS}`: e.g., "[skill-name]-专属进化规则.md, [skill-name]-最佳实践.md"
+
+---