You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@@ -97,6 +97,7 @@ Detailed rules are in the domain documents. Summary of responsibility:
97
97
-**Anti-monolith and code hygiene**: before creating or extending large modules, components, routes, services, prompts, or catch-all utility files, load `FCVW/skills/anti-monolith-guard/SKILL.md`. When duplication, stale files, dead code, repeated snippets, or retroactive cleanup are involved, load `FCVW/skills/code-hygiene-refactor/SKILL.md`. If a gate fails, split the plan before editing.
98
98
-**Agent/skill factory**: do not create new skills, agent profiles, specialist personas, or command packs by convenience. Load `FCVW/skills/agent-factory/SKILL.md`, prove recurrence, coverage gap, token/risk ROI, narrow scope, and validation before adding the asset.
99
99
-**Skill/agent self-improvement**: do not adjust existing skills or agent profiles for style-only preferences. Load `FCVW/skills/self-improvement/SKILL.md` and record evidence, metric passed, scope preservation, and validation replay before editing any skill/profile.
100
+
-**Wiki continuous learning**: when knowledge must be promoted, revised, grouped, tagged, linked, or scheduled for review, load `FCVW/skills/wiki-curator/SKILL.md`. Use the fixed optimized curation mode; do not expose customizable cost modes.
100
101
-**Documentation**: plans go in `FCVW/Plans/{status}`; `AGENTS.md` must be updated when new official documents are created; templates in `FCVW/governance/` follow structural changes of the VibeWork FrameCode.
101
102
-**Application module documentation**: when a change affects relevant pages, screens, modules, components, flows, or business rules in a downstream application, consult `FCVW/APPLICATION_DOCUMENTATION.md` and update the application-owned documentation path defined there.
102
103
-**Agent journals**: agent-specific journals must live under `FCVW/wiki/agents/<agent_name>_journal.md`; do not create competing journal paths.
@@ -134,7 +135,7 @@ Before executing a request that might modify files:
134
135
- for release, consult `FCVW/RELEASE.md`;
135
136
- for security, consult `FCVW/SECURITY.md`;
136
137
- for persistence, consult `FCVW/DATA.md`;
137
-
- for AI, RAG, memory, or continuous learning, consult `FCVW/AI.md`and `FCVW/wiki/schema.md`;
138
+
- for AI, RAG, memory, or continuous learning, consult `FCVW/AI.md`, `FCVW/wiki/schema.md`, and `FCVW/skills/wiki-curator/SKILL.md` when wiki curation, knowledge promotion, grouping, tagging, thematic colors, or review metrics are involved;
138
139
- for refactoring, consult `FCVW/REFACTORING.md`;
139
140
- for architectural decisions, consult `FCVW/ARCHITECTURAL_DECISIONS.md`;
140
141
- for auditing or release closure, consult `FCVW/AUDIT.md`;
Copy file name to clipboardExpand all lines: FCVW/AI.md
+10Lines changed: 10 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -66,6 +66,11 @@ Rules:
66
66
- avoid overwriting knowledge without backup;
67
67
- mark AI-generated content when applicable;
68
68
- allow human review when the content is critical.
69
+
- use `skills/wiki-curator/SKILL.md` when knowledge must be promoted, revised, grouped, deduplicated, tagged, or scheduled for review;
70
+
- prefer updating or superseding an existing wiki page before creating a new one;
71
+
- track freshness, promotion precision, duplication, release synthesis coverage, taxonomy coverage, and cost control using `wiki/metrics.md`;
72
+
- organize curated wiki pages with canonical tags, `theme`, and `theme_color` from `wiki/taxonomy.md`;
73
+
- use one fixed optimized curation mode only: JIT loading of index/log/schema/taxonomy/metrics plus directly triggered source records. Do not expose customizable curation cost modes.
69
74
70
75
### Agent with Tools
71
76
@@ -159,6 +164,7 @@ Rules:
159
164
- Learning generated from a conversation must be traceable.
160
165
- Agent-specific journals must use `wiki/agents/<agent_name>_journal.md`.
161
166
- Agent journals are append-only operational memory for durable project-specific learnings, not routine chat transcripts.
167
+
- Reusable journal learnings must be promoted or linked through `wiki-curator` instead of remaining isolated in chronological notes.
162
168
163
169
## AI Interaction Context Compression (AICC)
164
170
@@ -192,6 +198,7 @@ To deliver highly-specialized command sets and instruction procedures without in
192
198
-**Handoff Tracking:** Every skill file loaded during a session must be formally recorded under the `skills_invoked` section in the AICC Session Synthesis `S*.md` file.
193
199
-**Controlled Extension:** New skills and agent profiles are allowed only through `skills/agent-factory/SKILL.md`; convenience, naming, or persona-only additions are invalid.
194
200
-**Evidence-Based Self-Improvement:** Existing skills and agent profiles can be changed only through `skills/self-improvement/SKILL.md` when evidence proves failure, drift, validation gap, or meaningful token/risk reduction.
201
+
-**Wiki Curation:**`skills/wiki-curator/SKILL.md` owns continuous wiki learning tasks. Use it after releases, recurring troubleshooting, grouped notes, or explicit wiki-maintenance requests.
195
202
196
203
### 2. Agent and Skill Creation Metrics
197
204
@@ -293,6 +300,7 @@ To optimize execution speed, minimize financial API token costs, and prevent con
293
300
### 2. Context Boundaries & Pruning
294
301
* **Strict Domain Isolation:** Never open, read, or search files that are outside the active session type mapped in the table in `AGENTS.md` (e.g., in a bugfix session, do not open `DESIGN.md` or `DATA.md`).
295
302
* **Chunked View Limits:** Do not view entire large files. Limit reads using targeted line range parameters (`StartLine` and `EndLine`) to inspect only the required context.
303
+
* **Fixed Wiki Curation Mode:** When curating wiki knowledge, use the standard optimized mode in `skills/wiki-curator/SKILL.md`. Load only routing documents, wiki index/log/schema/taxonomy/metrics, and directly triggered source records unless `wiki-lint` finds an anomaly.
296
304
297
305
### 3. Log and Terminal Compaction
298
306
* **Silent Execution Flags:** When executing terminal commands, always use the shortest possible status flags (e.g., `git status -s` instead of `git status`) and suppress verbose outputs.
@@ -325,6 +333,8 @@ To facilitate retrieval and visualization in Obsidian, the AI must use the follo
325
333
- `#refactor-plan`: Plans and results of refactorings.
326
334
- `#user-feedback`: Insights and direct requests from the user.
327
335
336
+
Canonical themes, thematic colors, optional frontmatter fields, and extended tag guidance live in `wiki/taxonomy.md`. Freshness and curation metrics live in `wiki/metrics.md`.
337
+
328
338
## Models and Templates
329
339
330
340
To create new AI feature specifications, use the template in:
Copy file name to clipboardExpand all lines: FCVW/AUDIT.md
+4-1Lines changed: 4 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -32,6 +32,7 @@ Items:
32
32
- anti-monolith and code hygiene gates present when plans create or expand non-trivial modules;
33
33
- agent/skill creation gates present when plans add new skills, agent profiles, specialist roles, or reusable operational procedures;
34
34
- self-improvement gates present when plans modify skills, agent profiles, triggers, or agent operating rules;
35
+
- wiki-curation evidence present when plans promote, revise, group, retag, or schedule wiki knowledge;
35
36
- no plan closes with unresolved duplication, stale files, or dead-code cleanup claims without evidence or deferral;
36
37
- absence of contradictory instructions;
37
38
- updated templates.
@@ -112,6 +113,7 @@ Items:
112
113
- retrieved context;
113
114
- boundaries of action;
114
115
- memory and learning;
116
+
- wiki curation cost mode, taxonomy, freshness metrics, and source coverage;
115
117
- sensitive data;
116
118
- conflicting instructions.
117
119
@@ -134,6 +136,7 @@ Items:
134
136
-[ ] Plans that created skills, agent profiles, command packs, or reusable operational procedures recorded recurrence, coverage gap, token/risk ROI, scope boundary, and validation task.
135
137
-[ ] Plans that changed skills, agent profiles, triggers, or agent operating rules recorded failure/drift evidence, metric passed, scope preservation, token/risk ROI, and validation replay.
136
138
-[ ]`skills/README.md`, `CONTEXT_MAP.md`, and `STACK.md` list every active skill/profile exactly once.
139
+
-[ ] Wiki curation changes use `skills/wiki-curator/SKILL.md`, `wiki/taxonomy.md`, and `wiki/metrics.md` when tags, themes, grouping, freshness, or promotion are affected.
137
140
-[ ] No new agent or skill is persona-only, style-only, broader than its trigger family, or overlapping another asset by more than 50%.
138
141
-[ ]`R4` and `R5` plans include rollback and expanded validation, and `R5` plans record explicit human approval.
139
142
-[ ] High-risk low-priority plans were postponed, decomposed, discontinued, or explicitly justified.
@@ -147,7 +150,7 @@ Items:
147
150
-[ ]`SECURITY.md` reflects security changes.
148
151
-[ ]`AI.md` reflects AI changes.
149
152
-[ ]`TESTS.md` was used for validation compatible with risk.
150
-
-[ ] The governance wiki was updated when the audit generated reusable learning.
153
+
-[ ] The governance wiki was updated when the audit generated reusable learning, and `wiki/index.md` / `wiki/log.md` were synchronized.
151
154
-[ ] There are no temporary files being used as official source.
0 commit comments