원클릭으로
mirra-skills
Use Mirra to manage reusable instruction bundles for agent execution. Covers all Skills SDK operations via REST API.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Use Mirra to manage reusable instruction bundles for agent execution. Covers all Skills SDK operations via REST API.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use Mirra to living dashboards — natively-rendered grids of typed widgets (stat, image_card, list, progress, sparkline) that flows keep current by pushing data into them..... Covers all Dashboards SDK operations via REST API.
Use Mirra to execute user-defined scripts in aws lambda. Covers all Scripts SDK operations via REST API.
Use Mirra to the places a space publishes to — its corporate x, blog, newsletter — and the drafted copy waiting to go out to them. use listchannels to see what this space.... Covers all Space Channels SDK operations via REST API.
Use Mirra to the space's shared work-ledger. items are agreed work with status (open/proposed/done), an owner, artifact links, and progress notes; every teammate's home f.... Covers all Work Items SDK operations via REST API.
The team work-ledger ritual for agents on a Mirra space: track agreed work, propose discoveries (then ask in chat), relay approvals, close what ships, and publish ONE narrated update card per work burst — revise, never stack. Rides the Mirra items adapter / MCP work-ledger tools.
START HERE for anything Mirra. Load this whenever the repo you're working in has a .mirra/ directory (it's linked to a Mirra team space), or your human mentions their Mirra space, teammates' updates, or the team ledger. Directs the ambient team rituals — record work in the shared ledger, publish update cards, ask the space before expanding scope — and indexes every detail-level mirra-* skill.
| name | mirra-skills |
| description | Use Mirra to manage reusable instruction bundles for agent execution. Covers all Skills SDK operations via REST API. |
| allowed-tools | Read, Bash(curl:*, jq:*) |
Manage reusable instruction bundles for agent execution
You need the user's API key. Ask for these if not provided:
API_KEY: Mirra API key (generated in Mirra app > Settings > API Keys)API_URL: Defaults to https://api.fxn.world (only ask if they mention a custom server)All operations use a single POST endpoint with the resource ID and method in the body:
curl -s -X POST "${API_URL}/api/sdk/v2/resources/call" \
-H "Content-Type: application/json" \
-H "x-api-key: ${API_KEY}" \
-d '{
"resourceId": "skills",
"method": "{operation}",
"params": { ...args }
}' | jq .
Replace {operation} with the operation name from the table below.
Legacy alternative:
POST ${API_URL}/api/sdk/v1/skills/{operation}with args as the request body also works but is not recommended for new integrations.
| Operation | Description |
|---|---|
listSkills | List skills available to the user (built-in + user-created). |
getSkill | Get full details of a skill by ID. |
createSkill | Create a new user skill. |
updateSkill | Update a user skill procedure, templates, or metadata. |
deleteSkill | Delete a user skill. Cannot delete built-in skills. |
selectSkills | Select relevant skills for an agent iteration. Returns ranked skills with full procedure text. |
recordSkillUsage | Record that a skill was used during agent execution. |
getSkillUsage | Get usage records for a skill or flow. |
listSkillsList skills available to the user (built-in + user-created).
Arguments:
category (string, optional): Filter by categoryReturns:
AdapterOperationResult: Returns { count, skills[] }. Each skill: { skillId, name, title, description, category, builtIn, usageCount }.
Example:
curl -s -X POST "${API_URL}/api/sdk/v2/resources/call" \
-H "Content-Type: application/json" \
-H "x-api-key: ${API_KEY}" \
-d '{"resourceId":"skills","method":"listSkills","params":{}}' | jq .
getSkillGet full details of a skill by ID.
Arguments:
skillId (string, required): ID of the skillReturns:
AdapterOperationResult: Returns full skill with procedure, templates, and examples.
Example:
curl -s -X POST "${API_URL}/api/sdk/v2/resources/call" \
-H "Content-Type: application/json" \
-H "x-api-key: ${API_KEY}" \
-d '{"resourceId":"skills","method":"getSkill","params":{"skillId":"<ID>"}}' | jq .
createSkillCreate a new user skill.
Arguments:
name (string, required): Unique kebab-case nametitle (string, required): Human-readable titledescription (string, required): What this skill doeswhenToUse (string, required): Conditions for using this skillwhenNotToUse (string, required): Conditions to avoid this skillprocedure (array, required): Step-by-step procedurecategory (string, required): Skill categorytags (array, optional): Tags for discoveryReturns:
AdapterOperationResult: Returns the created skill.
Example:
curl -s -X POST "${API_URL}/api/sdk/v2/resources/call" \
-H "Content-Type: application/json" \
-H "x-api-key: ${API_KEY}" \
-d '{"resourceId":"skills","method":"createSkill","params":{"name":"<value>","title":"<value>","description":"<value>","whenToUse":"<value>","whenNotToUse":"<value>","procedure":[],"category":"<value>"}}' | jq .
updateSkillUpdate a user skill procedure, templates, or metadata.
Arguments:
skillId (string, required): ID of the skill to updateupdates (object, required): Fields to updateReturns:
AdapterOperationResult: Returns the updated skill.
Example:
curl -s -X POST "${API_URL}/api/sdk/v2/resources/call" \
-H "Content-Type: application/json" \
-H "x-api-key: ${API_KEY}" \
-d '{"resourceId":"skills","method":"updateSkill","params":{"skillId":"<ID>","updates":{}}}' | jq .
deleteSkillDelete a user skill. Cannot delete built-in skills.
Arguments:
skillId (string, required): ID of the skill to deleteReturns:
AdapterOperationResult: Returns { success, deletedAt }.
Example:
curl -s -X POST "${API_URL}/api/sdk/v2/resources/call" \
-H "Content-Type: application/json" \
-H "x-api-key: ${API_KEY}" \
-d '{"resourceId":"skills","method":"deleteSkill","params":{"skillId":"<ID>"}}' | jq .
selectSkillsSelect relevant skills for an agent iteration. Returns ranked skills with full procedure text.
Arguments:
agentObjective (string, required): The objective being pursuedcurrentPhase (string, optional): Current execution phaseisLooping (boolean, optional): Whether loop detection is activebudgetPercentage (number, optional): Token budget used (0-100)recentFailureCount (number, optional): Number of recent failureshasDelegation (boolean, optional): Whether delegation is configuredReturns:
AdapterOperationResult: Returns { count, skills[] } with full procedure text for LLM context injection. Max 10 skills.
Example:
curl -s -X POST "${API_URL}/api/sdk/v2/resources/call" \
-H "Content-Type: application/json" \
-H "x-api-key: ${API_KEY}" \
-d '{"resourceId":"skills","method":"selectSkills","params":{"agentObjective":"<value>"}}' | jq .
recordSkillUsageRecord that a skill was used during agent execution.
Arguments:
skillId (string, required): ID of the skill usedflowId (string, required): ID of the agent flowiteration (number, required): Iteration numberoutcome (string, required): success or failurenotes (string, optional): Optional notesReturns:
AdapterOperationResult: Returns the created usage record.
Example:
curl -s -X POST "${API_URL}/api/sdk/v2/resources/call" \
-H "Content-Type: application/json" \
-H "x-api-key: ${API_KEY}" \
-d '{"resourceId":"skills","method":"recordSkillUsage","params":{"skillId":"<ID>","flowId":"<ID>","iteration":10,"outcome":"<value>"}}' | jq .
getSkillUsageGet usage records for a skill or flow.
Arguments:
skillId (string, optional): Filter by skill IDflowId (string, optional): Filter by flow IDlimit (number, optional): Max records to return (default 20)Returns:
AdapterOperationResult: Returns { count, usageRecords[] }.
Example:
curl -s -X POST "${API_URL}/api/sdk/v2/resources/call" \
-H "Content-Type: application/json" \
-H "x-api-key: ${API_KEY}" \
-d '{"resourceId":"skills","method":"getSkillUsage","params":{}}' | jq .
All SDK responses return the operation payload wrapped in a standard envelope:
{
"success": true,
"data": { ... }
}
The data field contains the operation result. Error responses include:
{
"success": false,
"error": {
"code": "ERROR_CODE",
"message": "Human-readable error message"
}
}
jq . to pretty-print responses, jq .data to extract just the payloaddata.results or directly in data (check examples)--fail-with-body to curl to see error details on HTTP failuresexport API_KEY="your-key"