Tamar Resume Tailoring

Use the tamar CLI to tailor resumes for specific job applications via the Tamar API.

Triggers

Activate when the user says anything like:

• "tailor my resume"
• "customize my resume for"
• "help me apply for this job"
• "make a resume for this role"
• "adapt my resume"
• "target my resume at"

Prerequisites

• The tamar CLI must be installed by the user before using this skill: npm install -g tamar-cli
• An API key must be configured. Verify by running tamar status. If it fails with "No API key configured", ask the user to run tamar auth --key (keys are obtained from https://ask-tamar.com → Profile → API Keys)
• Do NOT read or inspect ~/.tamarrc directly — use tamar status to check auth

Pipeline

1. Check if the user has an experience profile

tamar profile

Shows existing profiles with name, role, seniority, skills, and enrichment depth. If a profile exists, use its ID with tamar tailor --profile for higher-quality output. If none exists, proceed to step 2.

2. Ensure user has a resume uploaded

If the user has a resume file and hasn't uploaded one yet:

tamar upload <resume-file>

3. Get the job description

Ask the user for the job description. It can be:

• A URL (LinkedIn, company careers page, etc.)
• A file path — read the file content first
• Pasted text

4. Tailor the resume

IMPORTANT — Input safety: Never interpolate user-provided strings directly into shell commands. Always write job descriptions or multi-word arguments to a temporary file and pass the file path, or use the -- separator and single-quote the argument. This prevents shell injection from malicious input.

# Safe: write JD to a temp file, pass the file
echo '<job description text>' > /tmp/jd.txt
tamar tailor --job /tmp/jd.txt

# Safe: single-quote the argument to prevent shell expansion
tamar tailor --job 'https://example.com/jobs/12345'

If the user also provides a resume file and hasn't uploaded before:

tamar tailor --job /tmp/jd.txt --resume '<resume-file>'

5. Review the output

The command returns JSON with:

• id — the generated resume ID (stored for later commands)
• quality — "enriched" (has profile) or "basic" (resume-only)
• analysis — job match analysis
• changes — list of changes made

Present the analysis summary conversationally. Highlight key matches and gaps.

6. Handle feedback

If the user wants changes:

Write the feedback to a temp file to avoid shell injection:

echo '<feedback text>' > /tmp/feedback.txt
tamar feedback "$(cat /tmp/feedback.txt)"

Or for a specific resume:

tamar feedback "$(cat /tmp/feedback.txt)" --id '<resume-id>'

7. Download the result

tamar download                    # PDF (default)
tamar download --format json      # structured JSON

The PDF is saved to the current directory. Tell the user the file path.

Error Handling

Quality Notes

• Enriched quality (user has an experience profile — check via tamar profile) = higher quality tailoring
• Basic quality (resume-only, no profile) = still useful but less nuanced
• If user has no profile, suggest building one at https://ask-tamar.com via the interactive Q&A, or use the enrichment API flow

Example Interaction

User: Can you tailor my resume for this job? https://linkedin.com/jobs/12345

Agent: Let me tailor your resume for that role.

[runs: tamar tailor --job 'https://linkedin.com/jobs/12345']

Agent: Done! Here's what I found:
- Quality: enriched (used your experience profile)
- Key alignments: Python, data pipelines, team leadership
- Adjusted: Reframed your experience to emphasize data platform work
- Gaps: Kubernetes — no production experience listed

Want me to tweak anything? I can also download the PDF for you.