Generate skill descriptions¶
This endpoint runs only the description-generation sub-step used by
generate_skills. It does not generate, normalize or filter any skills.
It uses AI to suggest missing descriptions for skills in an existing catalog. It attempts to generate one description for every unique received skill label, then reuses it for duplicates. The caller is responsible for filtering out skills whose description is already populated.
Request from Lexi¶
curl 'http://localhost:8000/lexi/api/v1/jobs/get_skill_descriptions' \
--header 'X-API-Key: <api-key>' \
--header 'Content-Type: application/json' \
--data '{
"skills": [
{
"skill_code": "SK-1",
"skill_title": "Python",
"skill_group": "TECHNIQUE"
}
],
"outputLanguage": "en",
"typeLLM": "openAI",
"debug": false
}'
{
"skills": [
{
"skill_code": "SK-1",
"skill_title": "Python",
"skill_group": "TECHNIQUE"
}
],
"outputLanguage": "en",
"typeLLM": "openAI",
"debug": false
}
outputLanguage defaults to en. typeLLM accepts openAI or Anthropic
and defaults to openAI. skill_code and skill_title are required.
Additional fields on skills are preserved in the response.
Processing logic¶
The skills are passed to the same function as the “Skill description” sub-step
of generate_skills, with the same options:
- global deduplication of skill labels;
- one generated description per unique skill;
- reuse of that description for all relevant jobs;
- the same description language and provider;
- batch generation with the same concurrency configuration.
Response¶
The response preserves the received skills and adds skill_definition:
{
"skills": [
{
"skill_code": "SK-1",
"skill_title": "Python",
"skill_group": "TECHNIQUE",
"skill_definition": "Ability to develop and maintain Python applications in order to deliver reliable software."
}
]
}
Generation errors return 400 with
{"detail":"Error during skills information retrieval"}.
The generated lexi HTTP reference is the authoritative source for parameters, schemas, responses and examples.