Loading...
Loading...
Found 3 Skills
Observe Before Editing
Reasoning protocol distilled from Claude Fable 5 — Floor check, multi-hypothesis diagnosis, adversarial self-review, Constraint Loop for hard output constraints. Use when the user says "fable5", "bật thinking mode", "reason like fable 5", or invokes /fable5. Once invoked it stays active for the rest of the session (toggle off with "stop fable5" / "normal mode").
Meta-skill for open-ended, multi-step work in any agent tool. Use when the request is vague, large, or spans research, planning, building, and writing. Routes the work through Clarify -> Plan -> Do -> Verify -> Humanize -> Deliver, instead of guessing. Do NOT use for a single narrow task that already has a specific skill (for example Firebase, UI audit, or codebase analysis) — use that skill instead, then run only the Humanize + Delivery Gate from this skill.