apply

/apply — End-to-End Application Pipeline

Purpose

Fire-and-forget pipeline that runs the full application workflow from company research through final deliverables. Produces a tailored resume (PDF/DOCX/MD/HTML), cover letter, scores, and review — all in one shot.

Optimized for multi-job-same-company use: company research is performed once and reused across subsequent runs.

Anti-Fabrication Rules (MANDATORY)

1. All sub-skills enforce their own anti-fabrication rules — this pipeline does not relax them
2. The final /verify step is the safety net — if it fails, flag the output clearly
3. No fabrication is acceptable regardless of pipeline speed or convenience

Session Logging Contract (MANDATORY)

Each step performs its sub-skill's analytical work inline. Every LOG GATE requires TWO writes before you may proceed:

1. Session YAML — the structured session file per the sub-skill's SKILL.md schema
2. Summary append — read the sub-skill's SKILL.md, find its "Append to summary.md" template, and execute it exactly

Both writes are gates. Do not proceed to the next step until BOTH are on disk. If you skip the summary append and only write the YAML, you have violated this contract.

Session files created during this pipeline:

1. knowledge/sessions/{slug}/{date}_research.yaml (if research performed)
2. knowledge/sessions/{slug}/{role-slug}/{date}_match.yaml
3. knowledge/sessions/{slug}/{role-slug}/{date}_qualify.yaml
4. knowledge/sessions/{slug}/{role-slug}/{date}_tailor.yaml
5. knowledge/sessions/{slug}/{role-slug}/{date}_score.yaml
6. knowledge/sessions/{slug}/{role-slug}/{date}_review.yaml
7. knowledge/sessions/{slug}/{role-slug}/{date}_cover-letter.yaml
8. knowledge/sessions/{slug}/{role-slug}/{date}_apply.yaml (aggregate, last)

Process

Step 1 — Research (conditional)

Extract the company name from the JD or user input. Check knowledge/sessions/{slug}/company.yaml for an existing CompanyProfile.

• If profile exists: Skip research. Log: "Reusing existing CompanyProfile for {company} (researched {date})."
• If no profile exists: Run /research with any links the user provided. Use non-interactive mode — do not ask for additional sources, work with what's given plus web search. LOG GATE: (1) Create knowledge/sessions/{slug}/{date}_research.yaml per /research Step 7 schema. (2) Read skills/research/SKILL.md → find the "Append to summary.md" template → append that filled template to knowledge/sessions/{slug}/summary.md. Do not proceed to Step 2 until both writes are on disk.

Step 2 — Match

Run /match against the JD. Do not prompt the user — proceed directly with wilson-resume.yml and the JD.

This produces: match scores, gap analysis, transferable skills, and tailoring recommendations.

LOG GATE: (1) Create knowledge/sessions/{slug}/{role-slug}/{date}_match.yaml per /match Step 5 schema. (2) Read skills/match/SKILL.md → find the "Append to summary.md" template → append that filled template to knowledge/sessions/{slug}/{role-slug}/summary.md. Do not proceed to Step 3 until both writes are on disk.

Step 3 — Qualify

Run /qualify using the CompanyProfile and JD. Do not prompt the user for culture preferences — score culture_fit with available signals and note confidence as low if preferences are unknown.

This produces: dimensional scores, weighted average, and strategic brief.

LOG GATE: (1) Create knowledge/sessions/{slug}/{role-slug}/{date}_qualify.yaml per /qualify Step 5 schema. (2) Read skills/qualify/SKILL.md → find the "Append to summary.md" template → append that filled template to knowledge/sessions/{slug}/{role-slug}/summary.md. Do not proceed to Step 4 until both writes are on disk.

Step 4 — Tailor

Run /tailor against the JD. This internally:

• Uses the CompanyProfile pain points for prioritization
• Uses the match session for gap awareness
• Uses the qualify session for positioning
• Produces tailored wilson-resume.yml in knowledge/sessions/{slug}/{role-slug}/tailored/
• Generates PDF, DOCX, HTML, MD outputs
• Runs its own quick persona check

Do not prompt for tailoring preferences — use match/qualify data to make decisions.

LOG GATE: (1) Create knowledge/sessions/{slug}/{role-slug}/{date}_tailor.yaml per /tailor Step 7 schema. (2) Read skills/tailor/SKILL.md → find the "Append to summary.md" template → append that filled template to knowledge/sessions/{slug}/{role-slug}/summary.md. Do not proceed to Step 5 until both writes are on disk.

Step 5 — Score

Run /score against the JD using the tailored resume (knowledge/sessions/{slug}/{role-slug}/tailored/wilson-resume.yml).

This produces: ATS and HR scores with component breakdowns.

LOG GATE: (1) Create knowledge/sessions/{slug}/{role-slug}/{date}_score.yaml per /score Session Log schema. (2) Read skills/score/SKILL.md → find the "Append to summary.md" template → append that filled template to knowledge/sessions/{slug}/{role-slug}/summary.md. Do not proceed to Step 6 until both writes are on disk.

Step 6 — Review

Run /review with the JD as context, evaluating the tailored resume. All personas report findings.

This is report-only — findings do not feed back into tailoring.

LOG GATE: (1) Create knowledge/sessions/{slug}/{role-slug}/{date}_review.yaml per /review Step 4 schema. (2) Read skills/review/SKILL.md → find the "Append to summary.md" template → append that filled template to knowledge/sessions/{slug}/{role-slug}/summary.md. Do not proceed to Step 7 until both writes are on disk.

Step 7 — Cover Letter

Run /cover-letter using the JD. This internally:

• Uses the CompanyProfile for company-specific hooks
• Uses the match session for gap mitigation
• Uses the qualify session for positioning angle
• Auto-selects the strongest hook type based on available data
• Saves to knowledge/sessions/{slug}/{role-slug}/tailored/cover_letter.md

Do not prompt for hook preference or talking points — auto-select based on qualify scores:

• Pain-solution match >= 7 → problem-solver hook
• ROI potential >= 7 → impressive-achievement hook
• Otherwise → specific-company-knowledge hook (if CompanyProfile exists) or impressive-achievement hook

LOG GATE: (1) Create knowledge/sessions/{slug}/{role-slug}/{date}_cover-letter.yaml per /cover-letter Step 7 schema. (2) Read skills/cover-letter/SKILL.md → find the "Append to summary.md" template → append that filled template to knowledge/sessions/{slug}/{role-slug}/summary.md. Do not proceed to Step 8 until both writes are on disk.

Step 8 — Verify

Run /verify on the tailored resume (knowledge/sessions/{slug}/{role-slug}/tailored/wilson-resume.yml).

Report status: PASS or FAIL with details. Do not block output delivery on verification — report it alongside results.

LOG GATE: Read skills/verify/SKILL.md → find the "Append to summary.md" template → append that filled template to knowledge/sessions/{slug}/{role-slug}/summary.md. Write before proceeding to Step 9.

Step 9 — Validate and Log Aggregate (MANDATORY — do not present results until this step is complete)

Validation: Confirm these sub-session files exist on disk (created at LOG GATEs above). If any is missing, create it now:

• knowledge/sessions/{slug}/{date}_research.yaml (if research was performed)
• knowledge/sessions/{slug}/{role-slug}/{date}_match.yaml
• knowledge/sessions/{slug}/{role-slug}/{date}_qualify.yaml
• knowledge/sessions/{slug}/{role-slug}/{date}_tailor.yaml
• knowledge/sessions/{slug}/{role-slug}/{date}_score.yaml
• knowledge/sessions/{slug}/{role-slug}/{date}_review.yaml
• knowledge/sessions/{slug}/{role-slug}/{date}_cover-letter.yaml

Summary reconstruction: Overwrite knowledge/sessions/{slug}/{role-slug}/summary.md by rebuilding it deterministically from session data:

1. Write the header: # {Company Name} — {Role Title}
2. List all session YAML files in knowledge/sessions/{slug}/{role-slug}/ matching {date}_*.yaml (sorted by filename)
3. For each session file (excluding _apply.yaml): a. Read the session YAML to extract its data b. Read the corresponding sub-skill's SKILL.md (e.g., skills/match/SKILL.md for _match.yaml) c. Find the "Append to summary.md" template in that SKILL.md d. Fill the template with data from the session YAML e. Append the filled entry to summary.md
4. Also check for knowledge/sessions/{slug}/{date}_research.yaml — if it exists and the company-level summary.md doesn't have a research entry, rebuild knowledge/sessions/{slug}/summary.md using skills/research/SKILL.md's template
5. Append the final "apply — Pipeline Complete" closing entry (per template below)

This guarantees completeness — the summary is derived from the actual session files on disk, not from trusting that each inline LOG GATE append happened correctly.

Aggregate session: Save to knowledge/sessions/{slug}/{role-slug}/{date}_apply.yaml:

date: YYYY-MM-DD
type: apply
company: Company Name
slug: company-slug
role: Role Title
role_slug: role-slug
output_dir: knowledge/sessions/{slug}/{role-slug}/tailored/
research_skipped: true | false
company_profile: knowledge/sessions/{slug}/company.yaml
sub_sessions:
  research: session-filename or null
  match: session-filename
  qualify: session-filename
  tailor: session-filename
  score: session-filename
  review: session-filename
  cover_letter: session-filename
  verify: pass | fail
scores:
  match: { required: XX, preferred: XX, overall: XX }
  qualify: { weighted_average: X.X, recommendation: strong-pursue | pursue-with-caution | consider-passing }
  ats: XX
  hr: XX
verification: pass | fail
output_files:
  - knowledge/sessions/{slug}/{role-slug}/tailored/wilson-resume.yml
  - knowledge/sessions/{slug}/{role-slug}/tailored/wilson-resume.pdf
  - knowledge/sessions/{slug}/{role-slug}/tailored/wilson-resume.docx
  - knowledge/sessions/{slug}/{role-slug}/tailored/wilson-resume.html
  - knowledge/sessions/{slug}/{role-slug}/tailored/wilson-resume.md
  - knowledge/sessions/{slug}/{role-slug}/tailored/cover_letter.md

Append closing entry to knowledge/sessions/{slug}/{role-slug}/summary.md (by this point, summary.md already contains individual entries from each LOG GATE above — this is the final closing marker):

---

## {date} apply — Pipeline Complete

**Pipeline**: research({skipped?}) → match → qualify → tailor → score → review → cover-letter → verify({PASS|FAIL})
**Scores**: Match {overall}% | Qualify {weighted_avg}/10 ({recommendation}) | ATS {score}/100 | HR {score}/100
**Output**: knowledge/sessions/{slug}/{role-slug}/tailored/
**Bottom line**: {1-2 sentence synthesis — what this means for the application, key strength to lead with, key risk to prepare for}

Output

Present a consolidated summary only after Step 9 is complete:

## Application Complete: {Company} — {Role}

### Scores
| Metric | Score |
|--------|-------|
| Match (overall) | XX% |
| Qualify (weighted avg) | X.X/10 — {recommendation} |
| ATS | XX/100 |
| HR | XX/100 |

### Verification
{PASS | FAIL with details}

### Review Highlights
- Top strength: ...
- Top concern: ...
- Consensus score: X.X/10

### Deliverables
All files in `knowledge/sessions/{slug}/{role-slug}/tailored/`:
- wilson-resume.yml (tailored)
- wilson-resume.pdf
- wilson-resume.docx
- wilson-resume.html
- wilson-resume.md
- cover_letter.md

### Strategic Position
{One-paragraph summary from /qualify strategic brief — how to position in interviews}

Options

• First run for a company: /apply Acme Corp jd.txt https://acme.com/about https://acme.com/blog
• Subsequent jobs at same company: /apply Acme Corp jd_backend.txt (reuses CompanyProfile)
• Minimal: /apply "Company Name" path/to/jd.txt

Non-Interactive Mode

This skill runs all sub-skills in non-interactive mode. Sub-skills that normally prompt for user input instead use these defaults:

• /research: Work with provided links + web search, no source-gathering prompt
• /cover-letter: Auto-select hook type, no preference prompt
• /qualify: Score culture_fit with available signals, no preference prompt
• /tailor: Use match/qualify data for decisions, no tailoring preference prompt
• /review: Run all applicable personas without selection prompt