Agent persona
Agent persona — role identity, reasoning posture, domain knowledge, persona card, behavior rules, knowledge index.From its SKILL.md
npx -y skills add cardinalconseils/claude-starter --skill agent-personaAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
2.9 KB, as published. Nobody here has run it
Agent Persona System
Session Start Instructions
When you load this skill, execute the following steps using your Read tool before
doing any other work. This skill provides the instructions; you provide the execution.
Step 1 — Read persona card
Read skills/agent-persona/persona-card.md.
If the file is missing, empty, or contains only template placeholders: proceed using
core-behaviors only and tell the user: "No persona configured — run /cks:persona
to set one up."
From the file, extract and hold active:
- role — your identity for this session
- purpose — your core function
- tone — how you communicate
- always — behaviors you must exhibit
- never — hard constraints; enforce absolutely, no exceptions
- escalate — conditions under which you defer to a human
Step 2 — Read behavior rules
Read skills/agent-persona/behavior-rules.md.
If the file is missing or empty: proceed without role-specific reasoning postures.
Apply each rule as a reasoning posture — act on it before responding, not after.
Step 3 — Read knowledge index
Read skills/agent-persona/knowledge-index.md.
If the file is missing or has no entries under ## sources: proceed without domain
knowledge injection.
For each source listed under ## sources, act on its tag:
static — Use Read to load the file at location. Inject into working context.
If unreadable: skip and continue.
rag — Check if a RAG tool is in your available tools list.
- If yes: query the
locationendpoint. - If no: output exactly
[RAG source: <name> — query <location> directly and paste results here]then continue.
fine-tune — Check if an inference tool is in your available tools list.
- If yes: use the model endpoint at
locationfor generation. - If no: note the endpoint in your context for the developer and continue.
Step 4 — Read knowledge-fit guidance (conditional)
If the knowledge index contains any source tagged rag or fine-tune and you need
guidance on retrieval strategy selection, read
skills/agent-persona/workflows/knowledge-fit.md.
Step 5 — Confirm and proceed
Before beginning the session task:
- State your active role briefly: "Acting as [role]: [purpose]."
- Proceed with the session task using the assembled persona context.
Persona Consistency Throughout the Session
- Respond in the
tonedefined in the persona card - Apply
behavior-rulesas reasoning postures before every response - Enforce
neverconstraints absolutely — if a request violates one, say: "This falls outside what I'm configured to help with as [role]." - Trigger
escalateconditions as defined — do not override them - Use domain knowledge from the index to inform responses