apply-dashboard

name	apply-dashboard
description	Your apply-ready pipeline view with quick actions. Shows scored roles grouped by decision tier, supports filtering (by score, company, decision, status), and routes actions to the right skill. Say "dashboard" to see what to apply to, "dashboard skipped" for skipped roles with reasons, "dashboard Anthropic" for company-specific view. Accepts actions like "cover letter for #68", "applied to Kadence", "skip #72", "answer questions for Harvey Director".
triggers	["db","dashboard","apply dashboard","show pipeline","what should I apply to","show apply queue","show me jobs","show me what to apply to"]

Apply Dashboard — Career OS Skill

Purpose

The single entry point for your post-scoring apply workflow. After the job-match-scorer runs, this skill renders the apply-ready pipeline and routes quick actions to the right downstream skill — so you never need to remember which skill handles what.

Think of it as a thin dispatcher: it owns the view (via pipeline-query.py) and the routing (to apply-tracker, cover-letter, resume-engine, and application-qa). It does NOT contain business logic for resume generation, question answering, or file updates.

Output Format

Always start your response with:

━━━ Career OS: Apply Dashboard ━━━

How to Invoke

View Commands

• dashboard — full apply-ready pipeline (≥80%, non-terminal), grouped by tier
• dashboard skipped — only skipped roles with reasons (for auditing job-match-scorer decisions)
• dashboard --min-score 70 — lower the threshold to include CHECK DELTA
• dashboard Anthropic — all Anthropic roles regardless of score
• dashboard --batch latest — only the most recent scoring batch
• dashboard applied — show what's been applied to (tracking)

Action Commands (routed to downstream skills)

• cover letter for #68 or cover letter for Harvey Director → cover-letter
• resume for #75 or resume for Kadence → resume-engine
• answer questions for #68 or answer #68 → application-qa
• applied to #75 or applied to Kadence → apply-tracker
• skip #72 or skip Canva → apply-tracker

---

DATA ARCHITECTURE

Script Dependency

This skill depends on pipeline-query.py located at:

~/.career-os-state/scripts/pipeline-query.py

The script reads the match tracker and pipeline files and outputs filtered, sorted results. The dashboard skill runs this script and renders its output.

Files This Skill Reads (via script)

File	Path	What It Provides
Match Tracker	career-intelligence/projects/job-search/job-pipeline-match-tracker.json	All scored roles across all batches
Pipeline	career-intelligence/projects/job-search/job-pipeline.json	stage_data — Applied/Active cross-reference

Files This Skill Writes

None. This skill is read-only. All mutations are routed to downstream skills.

---

BEHAVIOR: Render Dashboard

Step 1: Determine Intent

Parse the user's command to extract:

• View filters: min-score, decision tier, company, batch, status
• Action intent: cover letter, resume, answer, applied, skip
• Role reference: #N number or fuzzy name string

If the command is purely a view (no action), proceed to Step 2. If the command includes an action, proceed to Step 4.

Step 2: Run Pipeline Query

Execute the script with the appropriate flags. Map user intent to CLI args:

User Says	Script Args
dashboard	(defaults: --min-score 80, exclude terminal)
dashboard skipped	--decision SKIP --min-score 0 --include-closed
dashboard --min-score 70	--min-score 70
dashboard Anthropic	--company Anthropic --min-score 0 --include-closed
dashboard --batch latest	--batch latest
dashboard applied	--status APPLIED --include-closed
dashboard all	--min-score 0 --include-closed

Run:

python3 ~/.career-os-state/scripts/pipeline-query.py [flags] \
  --tracker-path $CAREER_HOME/career-intelligence/projects/job-search/job-pipeline-match-tracker.json

Step 3: Render Output

Present the script's grouped table output directly. Add a footer with available actions:

━━━ Career OS: Apply Dashboard ━━━

[grouped table output from script]

━━━ Actions ━━━
→ "resume for #N"                         — customize resume (DOCX+PDF)
→ "cover letter for #N"                   — generate cover letter (DOCX+PDF, opt-in only)
→ "answer questions for #N"               — portal question answers
→ "applied to #N" or "skip #N"            — record outcome
→ "dashboard skipped" to audit skip decisions
→ "cruise control" to batch-apply queued roles (resume only; cover letters are opt-in per WO-044)

Step 4: Resolve Role Reference and Route Action

When the user's command includes an action:

1. Extract the reference: Look for #N pattern or company/role name

2. Resolve via script:

   python3 pipeline-query.py --lookup "68" --format json
   

3. If ambiguous (multiple matches): Present disambiguation:

   Which Harvey role?
   A. #68 Engineering Director NY (88%)
   B. #85 Director Core Product (89% — already applied)
   C. #90 EM Product (86% — already applied)
   

   Wait for user to select before routing.

4. Route to downstream skill with full context pre-loaded:

Action	Downstream Skill	Context Passed
cover letter for #N	cover-letter	Company, role, JD URL, resume track, score
resume for #N	resume-engine	Company, role, JD URL, resume track
answer questions for #N	application-qa	Company, role, JD URL, score, match rationale
applied to #N	apply-tracker	Company, role, date (today), score
skip #N	apply-tracker	Company, role, reason (if user provides one)

When routing, provide the resolved context so the downstream skill does NOT need to re-read files or re-resolve the role:

The user wants a cover letter for:
- Company: Harvey AI
- Role: Engineering Director (NY)
- Score: 88% (APPLY tier)
- Resume Track: Exec
- JD URL: https://jobs.ashbyhq.com/harvey/3d3aaf03-...
- Warm Path: Cold

Please generate a cover letter following the cover-letter workflow.

---

SCORE QUALITY HANDLING

Roles with ⚠️ title-only Score Quality are treated specially:

1. Dashboard display: Show with ⚠️ prefix and warning:

   ⚠️ #75 Kadence VP Eng — 89% (title-only — re-score before applying)
   

2. Action suppression: Do NOT route cover letter for, resume for, or answer questions for actions to title-only roles. Instead:

   ⚠️ #75 was scored from title only (no JD text). Re-score with full JD
   before generating application materials.
   → Say "verify queue" to fetch the JD, then "rescore #75"
   

3. Status/skip still allowed: applied to #75 and skip #75 are permitted even for title-only roles (user may have applied manually or wants to skip based on their own judgment).

4. 🔄 partial scores: Displayed with note but actions are allowed:

   🔄 #68 Harvey AI Eng Director — 88% (partial JD — consider re-scoring)
   

---

EDGE CASES

• Script not found: If pipeline-query.py doesn't exist at the expected path, fall back to manually reading the match tracker and filtering. This is token-expensive but functional. Suggest the user run the installer.

• Empty results: If no roles match the filter, say so and suggest relaxing the filter: "No roles found at ≥80%. Try dashboard --min-score 70 to see borderline roles, or dashboard skipped to audit skip decisions."

• Stale data: The script reads files on every invocation — it's always current. If the user says "I just scored new roles" and they don't appear, check that the job-match-scorer wrote to the match tracker.

Greenhouse Portal Indicator (WO-043, REQ-001)

When rendering a dashboard entry for a company in the Greenhouse list (see apply-tracker/SKILL.md for the authoritative list) with status Applied or later, append a 🌱 indicator and the portal verification link:

#68 Harvey AI Engineering Director — Applied 2026-04-01 🌱
    → Verify: my.greenhouse.io/applications/

The 🌱 emoji signals "Greenhouse-hosted — portal verification available." For non-Greenhouse companies, omit the indicator and link. Pipeline entries that already carry a verified: greenhouse metadata tag can additionally show a ✓ verified suffix after the indicator.

---

INTERACTION WITH OTHER SKILLS

Skill	Relationship
job-match-scorer	Upstream — job-match-scorer writes data that dashboard reads
apply-tracker	Downstream — dashboard routes status updates here
cover-letter	Downstream — dashboard routes cover letter requests here
resume-engine	Downstream — dashboard routes resume requests here
application-qa	Downstream — dashboard routes portal question requests here
cruise-control	Peer — CC reads the same data for batch execution
pipeline-sync	Peer — sync validates the files that dashboard reads