SuperAPI AI Directory

ingest

General & other

View repository

Updated 10-16 07:18

About this skill

/ingest — Ingest Project Artifacts into Resume

Purpose

Analyze real project artifacts (source code, GitHub repos, blog posts, demos) and propose verified resume updates. Every claim must trace back to a source artifact.

Anti-Fabrication Rules (MANDATORY)

  • NEVER invent metrics (performance numbers, cost savings, user counts) unless provided by the user
  • NEVER claim technologies that appear only as transitive dependencies
  • NEVER embellish scope beyond what the code demonstrates
  • When you INFER something (e.g., scale from load test config), explicitly ASK the user to confirm
  • Provenance is REQUIRED on every proposed bullet

Process

Step 1 — Gather Artifacts

Ask the user what they have:

  • Source code path (local directory)
  • GitHub repository URL
  • Blog post or write-up
  • YouTube talk/demo link
  • Architecture diagrams
  • Document exports (HTML, PDF, markdown — from Confluence, Notion, Google Docs, etc.)
  • Any metrics they can provide (users, requests/sec, cost savings, etc.)

Step 2 — Analyze Artifacts

Analyze the provided artifacts to extract tech stack, architecture patterns, scale metrics, contribution scope, and timeline. Cross-reference claims across artifact types when possible.

Step 3 — Extract Verifiable Facts

Categorize findings:

  • Technologies: Only include if meaningfully used (imported AND used in core logic, not just transitive)
  • Architecture: Patterns visible in the code (event-driven, serverless, microservices, etc.)
  • Scale metrics: ONLY what's provable from artifacts. Flag anything inferred.
  • Test coverage: Number of tests, types of tests
  • Unique aspects: What makes this project notable?

Step 4 — Draft Resume Bullets

Generate 3-5 bullet options per project using the X-Y-Z formula:

"Accomplished [X] as measured by [Y] by doing [Z]"

Each bullet must include:

- label: "Descriptive Label"
  text: "The bullet text"
  technologies: [list, of, verified, tech]
  metrics: ["only verified metrics"]
  provenance:
    source: "ingested:{date}_{project-name}"
    artifacts: ["github:user/repo", "blog:post-name.md"]
    verified: false  # Set to false until user confirms

Step 5 — Propose Updates

Present to the user:

  1. Proposed resume.yaml additions — new project entry and/or new role bullets
  2. Proposed skills.yaml additions — new technologies with role tags
  3. Flags — anything that needs user confirmation (inferred metrics, scope claims)
  4. Questions — ask about metrics you couldn't find in artifacts

Wait for user review and approval before making any changes.

Step 6 — Apply and Verify

After user approval:

  1. Update resume.yaml with approved content (set verified: true on confirmed items)
  2. Update skills.yaml if new technologies were approved
  3. Run resume-builder verify to confirm provenance integrity
  4. Log the session to knowledge/sessions/projects/{date}_ingest_{project-slug}.yaml

Session Log Format

Save to knowledge/sessions/projects/{date}_ingest_{project-slug}.yaml:

date: YYYY-MM-DD
type: ingest
project: project-name
project_slug: project-slug
artifacts_analyzed:
  - type: source_code
    path: /path/to/project
  - type: github
    url: https://github.com/user/repo
facts_extracted:
  technologies: [list]
  architecture: [patterns]
  metrics: [verified metrics]
bullets_proposed: 5
bullets_accepted: 3
bullets_rejected: 2
skills_added: [new skills]

Output Requirements

  • Present findings in a clear, structured format
  • Clearly separate VERIFIED facts from INFERRED claims
  • Always ask before modifying resume.yaml
  • Bold key technologies and metrics in proposed bullets