Skip to main content

SC Software Leverage Assessment

Deprecated 2026-09-01 — superseded by the SC Engineering Portfolio.

The portfolio answers this document's question — does a codebase earn revenue (G1) or remove founder hours (G2)? — as the Impact dimension of its IEUP score, across 37 repositories rather than the five assessed here. Its Shelved table carries a per-initiative reason, which is what this document's "dead or near-dead" list was for.

The findings that were not already in the portfolio were merged into initiative decision logs on 2026-09-01: the intake loop that does not close (ENG-001), the coverage-and-trigger gap and the corrected claim about scheduled jobs (ENG-004), the 1Password coverage figure (ENG-005), and the corrected sc-compose repository finding (ENG-014). Work repos outside the measured 37 are now listed in the portfolio's Scope section.

Two claims below are now false. symphony-flow was cloned to this machine on 2026-09-01, and sc-compose is a proper checkout rather than a detached working copy. Read the portfolio, not this file, for current state.

Purpose​

The SC Software Repository Inventory answers what exists. This document answers a different question: does a given codebase earn revenue (G1), remove founder hours from routine delivery (G2), or neither? Read it before commissioning new software, because the recurring finding is that the useful thing was already built and then left alone.

Boundary: the inventory is auto-generated from github-automator/config/repo-inventory.yaml and must not be hand-edited. This document is hand-maintained and links to it.

The inventory is stale. It is stamped 2026-01-31, 6.5 months old. Roughly 19 repos sit at tier: 5, kind: TBD with no purpose recorded — and those are precisely the ones assessed below. It lists repos GitHub has since archived, and it has no notion of whether a repo is cloned locally, which is why nothing flags that Tier 1 symphony-flow is absent from the working machine. Regenerating it (gh-auto export-docs after refreshing the YAML) is an open follow-up.

Scope of the survey​

38 local git repositories — 22 under work/software/, 2 under work/cowork-presentation/, 16 under personal/software/ — plus the symphonycore-org (21 repos), score-ra (35) and ra-personal (5) GitHub organisations. Surveyed 2026-08-15. All personal repos assess as neither; they are omitted below.

The five with real leverage​

1. sc-client-intake — G2, live, extend it​

The only repo that is live, CI-deployed and sits directly on the founder-time critical path. A 61-field onboarding form at intake.symphonycore.com, a signed per-client prefilled link, and a Cloudflare Pages Function writing straight into the client's GHL sub-account custom values — the values every snapshot template and workflow reads. src/fields.ts is a declared single source of truth with a 61/61 field-parity test. Deploy runs green on every push to master.

The loop it does not close: the 61 fields land in GHL, but nothing downstream reads them into delivery, and the 09-clients/ profile is still authored by hand. Closing that — intake submission producing a client profile and the ClickUp onboarding tasks — is the largest reduction in founder onboarding hours available from code that already exists.

2. claude-code-skills + the repo's slash commands — G2, the delegation substrate​

39 skills, synced to ~/.claude/skills, plus three commands in this repo (/audit-client-tickets, /audit-stale-tickets, /generate-client-reference). The commands are already written in the shape delegation needs: each backed by a versioned playbook, each read-only until approval, each naming its guardrails.

The gap is coverage and trigger. Three commands against roughly 90 SOPs, most of which are still prose a human executes — and nothing invokes any of them on a schedule. Across all 38 repos there are no Claude subagents, no hooks, and no scheduled GitHub Actions; every workflow is push- or dispatch-triggered, and the machine's only cron is an unrelated whisper monitor.

3. digital-business-card — G1 candidate, and the timing is unusual​

Deployed at card.symphonycore.com. Next.js 14 on Cloudflare via OpenNext, with Stripe subscription billing and GHL CRM integration implemented in code, 30 of 34 requirements Done per its own status report. CLAUDE.md states the target plainly: BNI members. Untouched for 172 days.

Read the caution before acting on this. TODO.md carries an unchecked "Operational Setup (Pre-Launch)" list — the D1 database is not created, the Cloudflare environment variables including STRIPE_PRICE_ID are not set, and four GHL workflows do not exist. The MVP was also deliberately revised down to "Validate the core value prop... No payment required. Free cards only."

So this is a lead-capture play into the $199–$397/mo core service, not instant MRR. What makes it worth recording is the calendar: three BNI meetings fall between 20 and 27 August. The remaining work is dashboard configuration rather than engineering, which makes it a good Claude-in-Chrome task.

It is a candidate, not a commitment, and it must not displace collections work. Substituting a fresh technical idea for an overdue revenue conversation is the documented failure pattern — see sc-execution-history.md.

4. content-conductor — G2, broad delivery-task removal​

The largest working automation asset: 22+ CLI commands plus 9 anonymisation commands, 450 of 547 PRD requirements Done, Dockerised, with a Tailscale-bound HTTP layer. Its commands map onto work currently done by hand — generate-logo-package --client <slug> against the brand-kit SOPs, process-image --platform instagram for social assets, md2mail, generate-presentation, extract-pdf/extract-web, csv-map for data migration.

No SOP in this repo invokes it. The SOPs describe manual steps. Wiring named SOPs to specific commands converts a built tool into removed hours without writing a feature.

5. sc-compose — G1/G2, productizable, currently broken​

Two finished conversational document-creation apps (prd-creator, ghl-prd-creator) that run as Claude.ai Artifacts — no install, no API key, no backend for the end user — with 33/33 requirements Done and a reusable scaffold/ for producing more. This is the closest existing analogue to the SOP Agent product brief, and the scaffold is the how-to-build-it part that brief lacks.

Two concrete problems: the local directory has no .git — it is a detached working copy of score-ra/sc-compose, with .github/, CLAUDE.md and a status report but no repository — and the remote has not been pushed since 2026-03-08.

Not on this machine​

symphony-flow is Tier 1, described as "workflow orchestration platform for employee and client onboarding automation", integrates GHL, Google, ClickUp and Slack, and was pushed 2026-08-14. It is the most actively developed G2 asset in the organisation and there is no local clone. Any G2 conversation that ignores it is incomplete.

Dead or near-dead, recorded so they are not re-proposed​

RepoState
enterprise-glossary-systemREADME + PRD only, 6 files, no code
sc-data-quality-engineOne commit, all components marked "Planned", empty __init__.py files
SC-Intellegence-EngineExplicitly archived 2026-02-18, migrated to symphony-platform
writing-style-frameworkDead since 2026-01-09
sc-glossary-appDormant 175 days
reference-docs-1Duplicate clone of reference-docs from a second owner, 5 files behind
cowork-presentation/symphony-html-engineWorking copy with no commits at all — everything staged, never committed, no remote

Operational findings worth acting on​

  • CI is failing on two repos. customer-kb (CI and deploy, failing since at least 2026-08-07) and device-deployments (failing since 2026-07-29). sc-client-intake and sc-infrastructure are green.
  • sc-finance-sync has 33 credentials at 0% 1Password coverage, per the github-automator credential census. symphony-flow sits at 73%.
  • github-automator is worth reviving before any cross-repo work. Dormant 110 days but functional: a 15-command CLI with a 1,115-line inventory covering ~55 repos, carrying tier, purpose, tech stack, integrations, ports and a per-repo credential census, plus reconcile and update-repo-inventory commands that diff the YAML against live GitHub. The schema is good; the data is stale.

Revision History​

DateVersionChange DescriptionAuthor
2026-08-151.0Initial assessment across 38 local repos and 3 GitHub organisations. Records the five with real G1/G2 leverage, the absence of any Claude subagent, hook or scheduled workflow anywhere, the missing symphony-flow clone, and the staleness of the auto-generated inventory.Symphony Core Systems Team