How to use the Orienting to an Inherited Seat agent skill

What the skill makes the agent do

An agent skill is a folder with a SKILL.md file that tells an agent when and how to behave in a certain situation. This one applies when an agent has been primed into an existing seat through a planned handover, meaning a different agent retired and handed over the seat plus its earned context.

The skill teaches the agent to treat the handover as its own situation, distinct from a fresh launch or from context-window compaction. It tells the agent to keep its own identity, to treat the handover packet as claims to verify rather than facts to trust, to refuse stale prompts aimed at the seat name, and to ask its predecessor questions when needed. It also tells the agent to leave its own feedback and notes for the next occupant.

• Inherit the seat's mission, durable evidence, authority boundaries and lessons, but keep the agent's own identity and session.
• Re-check the work queue first instead of trusting the packet's snapshot.
• Route to role-specific onboarding where relevant.
• Stream feedback and re-deposit earned knowledge for the next occupant.

Install the skill from the ZIP download

Download the ZIP for this skill from this page, then unzip it. You can place the skill folder in one of two locations. For a personal install available across projects, unzip it into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder.

You can also upload the skill through the Claude apps by going to Settings, then Capabilities, then Skills, and uploading it there.

• Personal: ~/.claude/skills/
• Project only: .claude/skills/ inside your project folder
• Claude apps: Settings → Capabilities → Skills

Trigger the skill with natural requests

You do not run the skill directly. The agent loads it when the request matches the skill's description. The skill is meant for moments when the agent has just taken over an existing seat through a planned handover and needs a world model of what happened.

Phrase requests around the situation itself. Examples include asking the agent to orient itself after taking over a seat from a retired agent, asking it to check the handover packet before acting on it, or asking it what it should verify first after an inherited handover.

• "I was just handed this seat from a retired agent. Orient me before you act."
• "Check the handover packet claims before you rely on them."
• "What should I verify first now that I've inherited this seat?"
• "Who can I ask about why the predecessor made this decision?"

Know what the skill is not for

This skill is not the mechanic for the retiring agent. That side is covered by the sibling skill retiring-and-inheriting-a-seat. It also does not handle unplanned crashes or context-window compaction. For those, the skill points to claude-compaction-restore and session-compaction-and-restore.

If the agent is in your own session that hit the context limit, use the compaction-restore skills instead of this one.

Safety notes: scripts and network

The skill's listed requirements are nothing. It does not run scripts and does not use the network, according to its listing. The guidance does reference commands such as rig queue list and rig ask --help, which come from the OpenRig tooling the skill describes. Check that tooling is available in your environment if you want the agent to use those commands.

Treat answers from a consulted predecessor as testimony and note that the consulted tenure is retired. The skill's changelog notes that a missing history record, an unavailable runtime, a timeout, or an exhausted context can prevent an answer, so diagnose the specific failure before assuming predecessor access is closed.

• Requires: nothing listed
• Runs scripts: no
• Uses network: no

Tips for getting good results

Ask the agent to re-check the queue first, since work can land during the handover window. Ask it to map what durable surfaces should exist before deep work, so it does not assume the packet describes the whole world.

The skill is marked provisional in its metadata, so treat its guidance as current practice that may change.

• Check the queue before acting on the packet.
• Use rig ask --help to see the predecessor lookup options.
• Ask the agent to name which packet claims it verified and which it did not.