AGENTS.md

Purpose

This document defines how AI assistants (“agents”) may read, interpret, and modify this repository.
The goal is to enable résumé tailoring and automation without ever exposing confidential, private, or proprietary information.

Agents are expected to follow these instructions strictly when reading from or writing to files in this repository.

Scope of Work

Agents may:

Generate concise résumé variants, summaries, or formatting conversions (Markdown → PDF/HTML)
Provide editing suggestions aligned with job descriptions or target roles
Maintain tone, accuracy, and professional polish
Propose patch diffs or edited file content; do not run git unless explicitly authorized by the human

Agents must never:

Reveal, regenerate, or reconstruct any private data (e.g., full address, phone number, internal project names, non-public metrics)
Output or publish raw internal notes, metrics, or role dossiers that contain company-sensitive or confidential content
Introduce or invent details about the user’s employment history or performance

Privacy and Confidentiality Rules

✅ Allowed Information

Publicly shareable résumé content (titles, roles, years, skills, generic project descriptions)
Quantified results that are not company-sensitive and do not imply internal metrics
Industry-standard technology stacks (e.g., C++, Python, SystemC, etc.)
Publicly verifiable career history already visible on LinkedIn or similar profiles

🚫 Prohibited Information

Personal identifiers beyond:
- Name
- Email (professional)
- City/Region
- LinkedIn or GitHub profile
No phone numbers, street addresses, or personal URLs not meant for public use
No internal company names, code names, project numbers, or proprietary technology details
No confidential metrics (revenue, sales, unreleased product data, client names)

File Handling Rules

```
src/resume.md
```
is publishable and may be rewritten by agents.
```
src/includes/
```
is publishable and may be referenced or edited.

Private Working Directory (never committed)

```
work/
```
at the repository root is a local, git-ignored scratch area for temporary notes, raw dumps, or potentially confidential source material.
Agents may read documents in
```
work/
```
only to extract or summarize safe, public‑ready résumé content.
Agents must never:
- Copy or quote raw
```
work/
```
  content verbatim into
```
src/
```
  files or outputs.
- Commit, list, or otherwise expose filenames or contents from
```
work/
```
  .
- Include confidential metrics, client names, internal code names, or unreleased data sourced from
```
work/
```
  .

Resume Output Guidelines

Always write concise, factual, truthful content.
Target two pages maximum (≈900–1200 words) for the main résumé.
Tone: Professional, direct, and outcome-driven — not self-promotional.
Formatting: Semantic Markdown.
No hallucinated content.

Safe Prompt Template (for résumé tailoring)

SYSTEM: You are a professional résumé editor. Follow the privacy and formatting rules in AGENTS.md.

USER INPUTS:
1. Target job description
2. Current canonical résumé (`src/resume.md`)
3. Optional contextual excerpts from `src/roles/` and `src/includes/`
  - Optional private inputs: selected excerpts from `work/` (never quoted verbatim; sanitize and generalize)
4. Constraints:
   - Keep ≤ 2 pages
   - Preserve factual accuracy
   - Do not introduce confidential details
   - Remove phone number and street address
   - Output valid Markdown ready for Pandoc

OUTPUT:
- Tailored résumé in Markdown format
- No extra commentary or non-Markdown text
- No git/branch/commit operations

Verification Checklist

Before committing any AI-generated résumé to the repository:

Check	Pass Criteria
✅ No PII	No phone, address, or non-public email
✅ No confidential data	No client/internal project names or numbers
✅ Proper format	Markdown only, ≤ 2 pages
✅ No `work/` leakage	No raw quotes, filenames, or sensitive metrics from `work/`
✅ Professional tone	Neutral, factual, impact-focused
✅ No unauthorized VCS	No branch/commit/push/tag attempted

Agent Responsibilities

Treat this repository as public.
Follow
```
DESIGN.md
```
for repository structure and workflow.
Annotate or summarize — do not fabricate.

Operational Mode and VCS Policy

Default: advisory and file-scoped. Provide proposed changes as diffs or full file contents.
Do not create branches, commits, pushes, tags, or PRs unless explicitly requested by the human.
Never modify git settings, credentials, or remotes.

Revision Control

Do not perform git operations unless explicitly instructed by the human.

If asked to make commits/branches, use a dedicated branch:

app/<company>-<role>

After review, tags may be created on request:

git tag YYYY-MM-<company>-<role>

Never merge to
```
main
```
or push without explicit human approval.

AGENTS.md

Purpose

Agents are expected to follow these instructions strictly when reading from or writing to files in this repository.

Scope of Work

Agents may:

Generate concise résumé variants, summaries, or formatting conversions (Markdown → PDF/HTML)
Provide editing suggestions aligned with job descriptions or target roles
Maintain tone, accuracy, and professional polish
Propose patch diffs or edited file content; do not run git unless explicitly authorized by the human

Agents must never:

Reveal, regenerate, or reconstruct any private data (e.g., full address, phone number, internal project names, non-public metrics)
Output or publish raw internal notes, metrics, or role dossiers that contain company-sensitive or confidential content
Introduce or invent details about the user’s employment history or performance

Privacy and Confidentiality Rules

✅ Allowed Information

Publicly shareable résumé content (titles, roles, years, skills, generic project descriptions)
Quantified results that are not company-sensitive and do not imply internal metrics
Industry-standard technology stacks (e.g., C++, Python, SystemC, etc.)
Publicly verifiable career history already visible on LinkedIn or similar profiles

🚫 Prohibited Information

Personal identifiers beyond:
- Name
- Email (professional)
- City/Region
- LinkedIn or GitHub profile
No phone numbers, street addresses, or personal URLs not meant for public use
No internal company names, code names, project numbers, or proprietary technology details
No confidential metrics (revenue, sales, unreleased product data, client names)

File Handling Rules

```
src/resume.md
```
is publishable and may be rewritten by agents.
```
src/includes/
```
is publishable and may be referenced or edited.

Private Working Directory (never committed)

```
work/
```
at the repository root is a local, git-ignored scratch area for temporary notes, raw dumps, or potentially confidential source material.
Agents may read documents in
```
work/
```
only to extract or summarize safe, public‑ready résumé content.
Agents must never:
- Copy or quote raw
```
work/
```
  content verbatim into
```
src/
```
  files or outputs.
- Commit, list, or otherwise expose filenames or contents from
```
work/
```
  .
- Include confidential metrics, client names, internal code names, or unreleased data sourced from
```
work/
```
  .

Resume Output Guidelines

Always write concise, factual, truthful content.
Target two pages maximum (≈900–1200 words) for the main résumé.
Tone: Professional, direct, and outcome-driven — not self-promotional.
Formatting: Semantic Markdown.
No hallucinated content.

Safe Prompt Template (for résumé tailoring)

SYSTEM: You are a professional résumé editor. Follow the privacy and formatting rules in AGENTS.md.

USER INPUTS:
1. Target job description
2. Current canonical résumé (`src/resume.md`)
3. Optional contextual excerpts from `src/roles/` and `src/includes/`
  - Optional private inputs: selected excerpts from `work/` (never quoted verbatim; sanitize and generalize)
4. Constraints:
   - Keep ≤ 2 pages
   - Preserve factual accuracy
   - Do not introduce confidential details
   - Remove phone number and street address
   - Output valid Markdown ready for Pandoc

OUTPUT:
- Tailored résumé in Markdown format
- No extra commentary or non-Markdown text
- No git/branch/commit operations

Verification Checklist

Before committing any AI-generated résumé to the repository:

Check	Pass Criteria
✅ No PII	No phone, address, or non-public email
✅ No confidential data	No client/internal project names or numbers
✅ Proper format	Markdown only, ≤ 2 pages
✅ No `work/` leakage	No raw quotes, filenames, or sensitive metrics from `work/`
✅ Professional tone	Neutral, factual, impact-focused
✅ No unauthorized VCS	No branch/commit/push/tag attempted

Agent Responsibilities

Treat this repository as public.
Follow
```
DESIGN.md
```
for repository structure and workflow.
Annotate or summarize — do not fabricate.

Operational Mode and VCS Policy

Default: advisory and file-scoped. Provide proposed changes as diffs or full file contents.
Do not create branches, commits, pushes, tags, or PRs unless explicitly requested by the human.
Never modify git settings, credentials, or remotes.

Revision Control

Do not perform git operations unless explicitly instructed by the human.

If asked to make commits/branches, use a dedicated branch:

app/<company>-<role>

After review, tags may be created on request:

git tag YYYY-MM-<company>-<role>

Never merge to
```
main
```
or push without explicit human approval.

AGENTS.md

AGENTS.md

Purpose

Scope of Work

Privacy and Confidentiality Rules

✅ Allowed Information

🚫 Prohibited Information

File Handling Rules

Private Working Directory (never committed)

Resume Output Guidelines

Safe Prompt Template (for résumé tailoring)

Verification Checklist

Agent Responsibilities

Operational Mode and VCS Policy

Revision Control

Related Skills

Markdown Converter

Nano Banana Pro

1password

AGENTS.md

Purpose

Scope of Work

Privacy and Confidentiality Rules

✅ Allowed Information

🚫 Prohibited Information

File Handling Rules

Private Working Directory (never committed)

Resume Output Guidelines

Safe Prompt Template (for résumé tailoring)

Verification Checklist

Agent Responsibilities

Operational Mode and VCS Policy

Revision Control