From career
Reconciliation pass. Reads ground-truth.md + crm/companies.md + crm/opportunities.md + crm/outreach.md and surfaces discrepancies. Never silently overwrites — every finding is a proposed delta the user reviews. Writes to <WORKING_FOLDER>/reconcile/<YYYY-MM-DD>-findings.md.
npx claudepluginhub danielrosehill/claude-code-plugins --plugin careerThis skill is limited to using the following tools:
Workspace state drifts: ground-truth says "currently working with X" but X's engagement closed two months ago; ground-truth says "looking for full-time" but recent outreach is 90% consulting. This skill surfaces the drift; the user owns the reconciliation.
Guides Next.js Cache Components and Partial Prerendering (PPR): 'use cache' directives, cacheLife(), cacheTag(), revalidateTag() for caching, invalidation, static/dynamic optimization. Auto-activates on cacheComponents: true.
Guides building MCP servers enabling LLMs to interact with external services via tools. Covers best practices, TypeScript/Node (MCP SDK), Python (FastMCP).
Share bugs, ideas, or general feedback.
Workspace state drifts: ground-truth says "currently working with X" but X's engagement closed two months ago; ground-truth says "looking for full-time" but recent outreach is 90% consulting. This skill surfaces the drift; the user owns the reconciliation.
$ARGUMENTS:
--strict — surface even minor signal mismatches.--apply — for findings the user pre-approves, apply changes (rare; default is propose-only)./career:reconcile
/career:reconcile --strict
ground-truth.md (all sections).crm/companies.md, crm/opportunities.md, crm/outreach.md.plans/.Each check has a description, the rule, and the action.
looking-for says "currently working with client X" but crm/opportunities.md shows that engagement at status closed-won 90+ days ago, or closed-lost, or paused → propose ground-truth update.hard-constraints excludes a region (e.g. US-only) but recent recommendations include US-only companies → flag the recommendation logic, not ground-truth.recommendations/ → flag mismatch.crm/companies.md rows with status engaged but no outreach.md activity in 60+ days → propose status change to paused or passed.crm/outreach.md rows in status sent for >45 days with no follow-up → propose status stalled.domains-of-interest lists 4+ → flag narrowness.domains-of-interest lists a domain with no domain-notes file → suggest running /career:discover by-domain for it.companions.transcription set in config but companion not actually installed → flag.freelance-brand in ground-truth is empty → flag.Tag each finding:
critical — internal contradiction (claim X is currently true, evidence X is false).drift — direction has shifted; ground-truth probably needs editing.staleness — row hasn't been touched, low cost to fix.coverage — gap (a domain not yet researched, a config not set).--strict surfaces signal mismatches that would otherwise be filtered (small drifts, single-row stale rows).
# Reconcile findings — {{date}}
> Generated: {{ISO timestamp}}
> Sources: ground-truth.md ({{mtime}}), crm/* ({{summary}}), recent recommendations + week-logs.
## Critical
### {{finding-id}}: {{title}}
- evidence: {{specific row + file}}
- rule: {{which check fired}}
- proposed action: {{e.g. "edit ground-truth section `looking-for`"}}
- to apply: `{{exact next command}}`
## Drift
(same shape)
## Staleness
(same shape)
## Coverage gaps
(same shape)
## No discrepancies found
(if a category is clean, list it here so the user sees the system checked)
If --apply and a finding has a clearly-mechanical action (e.g. flip outreach row from sent → stalled after 45 days):
Never auto-apply ground-truth edits — those always remain proposals.
findings: reconcile/<date>-findings.md
critical: <n>
drift: <n>
staleness: <n>
coverage: <n>
next:
critical → /career:ground-truth edit-section <section>
staleness → /career:reconcile --apply (mechanical updates only)
coverage → /career:discover by-domain "<missing>"
staleness to critical because the user "should" act on it.