eliasforge/fable-agent-orchestration/flag-safe-handoff
Claude Code and agent skill by eliasforge/fable-agent-orchestration
- community
Checking that the SKILL.md file is still in the repository.
Use when a delegated agent repeatedly flags a legitimate, authorized, mechanically benign task because domain vocabulary triggers a false positive. The conductor inspects the source and sends a narrow exact-edit packet so the worker does not need to survey the triggering modules. Never use this skill to conceal unsafe intent or bypass a valid refusal.
Install
npx vskill@latest install eliasforge/fable-agent-orchestration/flag-safe-handoffSKILL.md
Flag-Safe Handoff
Use this skill to route around a verified vocabulary false positive without weakening the task, hiding intent, or asking a worker to ignore a valid safety boundary.
Preconditions
All of these must be true:
- the work is authorized, legitimate, and mechanically benign;
- the conductor has read the triggering source and understands the requested
change;
- the refusal is caused by domain wording rather than the actual objective;
- the change can be bounded to exact files, signatures, behavior, and tests.
If any precondition is uncertain, stop and investigate. Use a normal handoff when no false positive has been demonstrated.
Procedure
- The conductor reads the relevant source and decides the exact change.
- Replace domain-flavored names with neutral opaque role labels in the packet.
Preserve exact signatures, types, limits, and old-to-new behavior.
- Name the files the worker owns and the triggering modules it must not survey.
- State the original invariant, explicit non-scope, broken behavior, expected
fixed evidence, and required gates.
- The worker implements only the supplied edits. It must not reconstruct,
broaden, or reinterpret the surrounding domain task.
- The conductor independently reads the resulting diff and runs the real
task-relative gate.
Packet Template
Acceptance Gate
The handoff passes only when:
- the exact original task advances;
- the worker stayed inside the supplied file and behavior boundary;
- the worker did not need to inspect the triggering modules;
- the conductor verified the actual diff rather than trusting the receipt;
- the task-relative gate fails under the old behavior and passes after the
change.
Invalid Uses
Do not use this skill to:
- conceal or euphemize a genuinely unsafe objective;
- ask a worker to bypass a valid safety restriction;
- omit material behavior from the packet;
- turn a broad or poorly understood task into blind edits;
- weaken tests because the original request was difficult to delegate.
If a precise benign packet is still rejected, use a different authorized execution surface or complete the change locally. Do not loosen the invariant.
Pairs well with agent-dispatch-packet, task-relative-test-gate, two-critic-review-loop, and review-verifier.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/flag-safe-handoff/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/eliasforge/fable-agent-orchestration/flag-safe-handoff)