Agent skill

get-output-build

Fetch VectorShift pipeline outputs by task_id (or span_id) and pass them to a builder skill

View SKILL.md on GitHub Repository
Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/get-output-build

SKILL.md

Get Output & Build

Fetches completed VectorShift pipeline outputs using task_id(s) and chains them into a builder skill for processing.

Note: task_id and span_id are interchangeable. VectorShift confirmed that the task_id returned when you submit a job can be used directly to query results.

Slash Command Usage

/get-output-build <task_id> --builder "<skill_command>"

Examples:

  • /get-output-build 6961e088862a01eeb682196b --builder "/process-ngm-lectures 1-OeTj2AseENWJFN-0jKz8tZqFDkLsFR2 --module-id module-1"
  • /get-output-build task1 task2 task3 --builder "/process-ngm-lectures 1ABC123 --module-id foundations"

Required Configuration

Environment Variables

bash
VECTORSHIFT_API_KEY=sk_6L1cRGLL5qN2d9rjViLVAFr6ATqE1OAE78L5bgYltWBkryoE

How It Works

Step 1: Fetch Pipeline Output

For each task_id, make a GET request to:

GET https://api.vectorshift.ai/v1/pipeline/6961308d6fdec16163ee0e2f/run/status/{task_id}
Authorization: Bearer {VECTORSHIFT_API_KEY}

Note: task_id (from job submission) works interchangeably with span_id for status queries.

Response (completed):

json
{
  "task_id": "...",
  "status": "completed",
  "result": {
    "content_analysis": "...",
    "lecture_json": "{ ... JSON lecture object ... }",
    "research_dossier": "...",
    "slide_blueprint": "...",
    "transcript": "..."
  }
}

Step 2: Extract Output Data

The pipeline returns multiple outputs. The key output for lecture building is lecture_json.

python
import json
import re

def extract_json(raw_output):
    """Extract JSON from VectorShift output (may be wrapped in markdown fences)"""
    content = raw_output.strip()
    if content.startswith('```json'):
        content = content[7:]
    elif content.startswith('```'):
        content = content[3:]
    if content.endswith('```'):
        content = content[:-3]
    
    match = re.search(r'\{[\s\S]*\}', content)
    if match:
        return json.loads(match.group())
    return None

Step 3: Pass to Builder Skill

The fetched outputs are passed to the builder skill command. The builder skill receives:

  • The extracted lecture_json data
  • Any additional parameters from the original command

Execution Workflow

User: /get-output-build abc123 def456 --builder "/process-ngm-lectures 1-OeTj --module-id module-1"

Step 1: Fetch outputs from VectorShift
  GET /v1/pipeline/.../run/status/abc123 → lecture data
  GET /v1/pipeline/.../run/status/def456 → lecture data

Step 2: Collect all lecture JSON outputs
  - Lecture 1: { "title": "...", "sections": [...] }
  - Lecture 2: { "title": "...", "sections": [...] }

Step 3: Pass to process-ngm-lectures
  - Save lecture JSONs to content/ngm-lectures/{module-id}/
  - Update registry.ts with new imports and entries

Python Script

Use the provided fetch_outputs.py script:

bash
python3 .claude/skills/get-output-build/fetch_outputs.py \
  --task-ids abc123 def456 ghi789 \
  --output-dir /tmp/vs-outputs

This creates JSON files for each task_id in the output directory.

Note: The script accepts both --task-ids and --span-ids (they're equivalent).

Error Handling

Status Action
completed Extract result and continue
in_progress Wait and retry (poll every 10s)
failed Log error and skip this task_id
HTTP 404 Invalid task_id, skip
HTTP 401 Invalid API key

Integration with process-ngm-lectures

When using with /process-ngm-lectures, the workflow becomes:

  1. You already have task_ids from a previous VectorShift run
  2. Run this skill to fetch the completed outputs
  3. Outputs are saved to content/ngm-lectures/{module-id}/
  4. Registry is updated automatically

Bypassing Google Drive

This skill is useful when:

  • Pipeline jobs were submitted earlier and you have the task_ids
  • You want to resume a failed run using cached task_ids
  • Pipeline outputs are ready but weren't processed

File Locations

File Purpose
.claude/skills/get-output-build/SKILL.md This documentation
.claude/skills/get-output-build/fetch_outputs.py Output fetching script
content/ngm-lectures/{module-id}/ Where lecture JSONs are saved
content/ngm-lectures/registry.ts Module/lecture registry

API Details

Property Value
Base URL https://api.vectorshift.ai/v1
Pipeline ID 6961308d6fdec16163ee0e2f
Status Endpoint GET /pipeline/{id}/run/status/{task_id}
Auth Authorization: Bearer {api_key}

Note: task_id and span_id are interchangeable in the status endpoint.

Example Complete Session

User: /get-output-build 6961e088862a01eeb682196b 6961e089862a01eeb682196c --builder "/process-ngm-lectures 1-OeTj2AseENWJFN-0jKz8tZqFDkLsFR2 --module-id module-1"

Claude: I'll fetch the VectorShift outputs and process them.

Step 1: Fetching outputs from VectorShift API...
  task_id 6961e088862a01eeb682196b: ✓ completed
    → Lecture: "Cellular Senescence and the SASP"
  task_id 6961e089862a01eeb682196c: ✓ completed
    → Lecture: "The SHIFT Framework"

Step 2: Extracting lecture JSON data...
  Extracted 2 lectures

Step 3: Executing builder command...
  Running: /process-ngm-lectures (skip to Phase 4)
  
  Saving lecture JSON files...
    Saved: content/ngm-lectures/module-1/cellular-senescence-and-the-sasp.json
    Saved: content/ngm-lectures/module-1/the-shift-framework.json
  
  Updating registry.ts...
    Added 2 imports
    Updated module-1 entry

Step 4: Preview URLs ready!
  http://localhost:3000/preview/ngm/module-1/cellular-senescence-and-the-sasp
  http://localhost:3000/preview/ngm/module-1/the-shift-framework

Recommended Agent Skills

Expand your agent's capabilities with these related and highly-rated skills.

majiayu000/claude-skill-registry

agent-ops-spec

Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-state

Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-spec

Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-testing

Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-testing

Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-state

Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.

163 31
Explore

Didn't find tool you were looking for?

Be as detailed as possible for better results