Cc concept onboarding

Onboarding Skill

This skill collects the strategic context downstream cc-concept skills need to run — positioning, competitive analysis, go-to-market planning, and campaign concepting. Context files are Markdown documents registered in the ## Context files table of CLAUDE.md. Skills load them on demand to set strategy accurately and at the right depth instead of defaulting to generic frameworks. This skill runs standalone and also recognizes context that an existing cc-content project already provides.

This skill follows the shared step sequence in ../_shared/skill-contract.md. Steps below add only onboarding-specific behavior. (Steps 5 and 6 from the contract are omitted because this skill collects multiple context files rather than emitting a single artifact.)

Step 0: Recall learnings

If .claude/learnings.md exists, read it silently. Apply relevant entries — in particular, path preferences, rejected interview answers, or project constraints discovered in prior runs. Do not announce this step. If the file is absent, continue normally.

Step 1: Discover existing context

This skill runs inside the owner's project — an arbitrary repo that may already have its own structure, possibly with more than one CLAUDE.md at different levels (root plus nested ones for sub-brands or campaigns). Never assume a single root-level file. Scan the whole project:

find . -name 'CLAUDE.md' -not -path '*/.git/*' -not -path '*/node_modules/*' | sort
find . -type d -iname 'context' -not -path '*/.git/*' -not -path '*/node_modules/*' | sort

For each context/-like folder found, list its files too — the ## Context files table only shows registered entries, and a folder can hold files nobody registered yet:

for d in <each context-like folder path from the find above>; do
  echo "=== $d ==="
  find "$d" -type f -name '*.md' | sort
done

For each CLAUDE.md found, check whether it already has a ## Context files table:

for f in <each CLAUDE.md path from the find above>; do
  echo "=== $f ==="
  grep -A 200 '## Context files' "$f" 2>/dev/null || echo "(no context table)"
done

Keep the full list of discovered CLAUDE.md paths — Step 4 needs it to decide where each newly registered file belongs. Show the owner the full picture, grouped by location:

"Here's what your project already has set up:

Root (CLAUDE.md)

Label	File	Summary
<existing rows, or "(none)">

(and nested locations if found)"

Step 2: Coverage check

For each gated need below, decide covered / uncovered by semantic match against the Summary column of the context table — never against a label or filename. The two needs are evaluated independently — one being covered does not exempt the other. Explicitly, if the project already has organization background covered (typical in a cc-content project), and competitive landscape uncovered (cc-content never collects that), then collect only competitive landscape.

Gated need: Organization background — the organization's mission, values, size, market position, customer segments, and constraints. Keywords in Summary: organization, company, business model, background, profile, audience, market position, customer, constraints, mission, values, team, size.

Gated need: Competitive landscape — the competitive set, key competitors and their positioning, market gaps, differentiation opportunities, and customer pain points. Keywords in Summary: competitor, competitive, competitive landscape, differentiation, market gap, positioning, competitive set, pain points.

For each uncovered need, proceed to Step 3. If all needs are covered, there is nothing to collect or register; go to Step 7 (Feedback).

Step 3: Collect

For each uncovered need, follow one of two paths:

Path A: Register a passed-in file. If $ARGUMENTS contains a file path, use that as the starting point for this need. Read it, confirm it covers the need, and continue to collect any remaining uncovered needs. All collected files (passed-in and interviewed) are registered together in Step 4.

Path B: Guided interview. For each uncovered need, run a brief interview to collect the information. After the interview, write the response to a file with a collision-checked name. Apply this collision-check logic:

1. Propose a default filename by convention (e.g., context/organization-background.md for the organization need, context/competitive-landscape.md for the competitive need).
2. Check if a file at that path already exists.
3. If it exists, generate a distinct filename by appending a -N suffix before the .md extension (e.g., context/competitive-landscape-2.md). Keep incrementing N (try -2, then -3, etc.) until a filename that does NOT already exist is found. Never overwrite an existing context file.
4. Write to the final, collision-checked filename.

Sample interview for organization background:

• Who is this organization? (name, type, size)
• What does it do? (products, services, market)
• Who does it serve? (customer segments, use cases)
• What are the business constraints? (budget, team size, timeline, channels)

Sample interview for competitive landscape:

• Who are the main competitors? (2–5 names + brief positioning)
• What are they doing well?
• What are the market gaps or underserved customer needs?
• How does this organization want to differentiate?

Step 4: Register

For each newly collected file, add a row to the ## Context files table in the appropriate CLAUDE.md (the root level, or the most specific nesting level that applies). Use this format:

| Label                   | File                                 | Summary                                                         |
| ----------------------- | ------------------------------------ | --------------------------------------------------------------- |
| Organization background | `context/organization-background.md` | Company mission, values, size, market position, and constraints |
| Competitive landscape   | `context/competitive-landscape.md`   | Key competitors, market gaps, and differentiation opportunities |

The Summary column must be semantic and describe the need it covers — use keywords that match the ones in Step 2 so future coverage checks will find it. Do not hand-edit the Key Config Files table — the pre-commit hook owns that sync.

Step 7: Feedback

Store learnings tagged [cc-concept:cc-concept-onboarding] in .claude/learnings.md. For example:

[cc-concept:cc-concept-onboarding] client prefers "go-to-market plan" over "GTM strategy" — 2026-07-17
[cc-concept:cc-concept-onboarding] org-background file should ask about primary channels early — 2026-07-17

If the session revealed a rejected answer, path preference, or project constraint that future runs should skip, record it now.