The system behind the product
TrustGrowth separates evidence collection, research, strategy, and analysis into specialist roles with visible handoffs. Groundcrew makes the operating instructions inspectable too.
curl -fsSL https://raw.githubusercontent.com/techwright-lab/groundcrew/main/install.sh | bash
Works with Claude Code, Cursor, Hermes, OpenClaw — the skills are plain Markdown.
Working today
These are current product capabilities. Examples describe TrustGrowth operating on its own site; they are dogfood evidence, not claims that TrustGrowth caused a growth outcome.
Crawls a site and records current technical, performance, content, growth, and trust findings with the affected pages attached.
Current dogfood example
On trustgrowth.ai, the Auditor produces the current crawl findings shown in the public proof baseline.
Collects keyword, competitor, authority, and search evidence where the connected account and plan make those sources available.
Current dogfood example
For TrustGrowth's own site, the Researcher maintains inspectable keyword opportunities alongside first-party Search Console evidence.
Turns approved audit and research evidence into a site-specific strategy that remains reviewable before downstream work begins.
Current dogfood example
TrustGrowth stores its own strategy as a review-gated artifact tied back to the evidence used to prepare it.
Compares current scores, issues, Search Console evidence, and recent changes without presenting correlation as causation.
Current dogfood example
The public TrustGrowth proof shows the Analyst's current baseline and what remains weak, rather than attributing movement to the product.
Joining soon
These roles are not presented as active. The descriptions define the reviewable artifact each role is intended to add, without promising an outcome or a release date.
A paced content calendar built from an approved strategy, with each proposed entry available for review.
Briefs and long-form drafts inside the review lifecycle; drafts do not publish themselves.
A documented check of factual support, claim risk, and approval state before a draft can move forward.
The public playbook
Groundcrew is TechWright's public pack of agent skills. We publish it so the instructions behind bounded growth and engineering workflows can be read, reused, and challenged instead of hidden behind an “AI-powered” label.
Skills encode repeatable tasks such as inspecting TrustGrowth evidence, triaging findings, and handing bounded work to the right tool or agent.
A role receives a narrow job, reads the relevant evidence surface, follows the skill's gates, and leaves an inspectable artifact or handoff. Human approval remains explicit where the workflow requires it.
The repository exposes the operating method. Buyers and builders can inspect the same playbook behind the roster rather than relying on a black-box team claim.