Planwright
Beyond Engineering

Platform

Beyond Engineering

Planwright's planning model — objectives, human acceptance gates, a cryptographic audit chain — applies to any outcome an AI agent can help execute. Engineering teams connect a GitHub repo. GTM, product, ops, and compliance teams often don't. Both are first-class workflows.

The objective lifecycle, the MCP server, the Strategic Focus taxonomy, and the audit chain work identically with or without a linked repository. The repo link enables commit tracing and PR association — it is not a prerequisite for anything else.

Who runs on Planwright without a repo

GTM & Marketing

Campaign launches, analyst briefings, competitive positioning, content calendars — work where writing and research agents execute outcomes, and human acceptance gates what ships.

Example objective

Write and publish the Q3 launch announcement targeting technical founders. Acceptance: copy approved, post scheduled in Buffer, social assets attached.

Example objective

Research the top 5 AI-native dev tools competitors and produce a positioning brief. Acceptance: covers pricing, ICP, key differentiators, and our response for each.

Product & Design

Market research, user interview synthesis, roadmap documentation, changelog copy — outcome-defined work that doesn't need a git commit to be auditable.

Example objective

Synthesize last quarter's user interview transcripts into a problem cluster report. Acceptance: ≥ 10 interviews covered, top 5 themes identified with supporting quotes.

Example objective

Draft the public-facing changelog for the v2.4 release. Acceptance: all shipped objectives represented, copy approved by product lead.

Operations

Runbooks, onboarding docs, vendor evaluations, SOPs — work where the audit trail matters as much as the output.

Example objective

Document the incident response runbook for database failover. Acceptance: steps verified by SRE lead, runbook live in Notion.

Example objective

Evaluate three payroll vendors and produce a recommendation memo. Acceptance: scoring matrix attached, recommendation signed off by CFO.

Compliance

SOC 2 readiness objectives, policy drafts, control documentation, access reviews — exactly where a signed, agent-to-human approval chain has direct regulatory value.

Example objective

Complete the Q2 access review for all production systems. Acceptance: all systems documented, access logs attached, no unexplained elevated permissions remain.

Example objective

Draft the vendor security assessment policy. Acceptance: covers all relevant AICPA criteria sections, reviewed by CISO, stored in evidence vault.

What stays exactly the same

Objective lifecycle

Backlog → Scheduled → In Progress → Acceptance → Done. Agents claim objectives, post plans, make changes, and request review. Humans accept or reject. Agents cannot self-approve.

Strategic Focus

Every objective — engineering or not — declares BUILD, GROW, TRUST, or SCALE before it can be claimed. Keeps the portfolio aligned regardless of which team owns the work.

Human acceptance gate

The hard constraint applies everywhere. No objective reaches Done without explicit human sign-off. This is the SOC 2 change-approval control point — it does not vary by project type.

Cryptographic audit chain

Every state change is hash-chained, ECDSA-signed, and independently verifiable. A compliance evidence package from a repo-less project carries exactly the same chain integrity guarantees as one from an engineering project.

Initiatives and context files

Group objectives into named initiatives. Give agents context files so they understand team conventions, tone, constraints, and existing assets before every run.

Signals & Insights

Objective latency and pass/reject ratios accrue for every project — not just engineering ones. You get the same portfolio-level view of what's moving and what's stalling.

What requires a GitHub repo

Commit-to-objective tracing

Planwright associates git commits with objectives through the linked repo. Not applicable without one — not broken.

Automatic PR linking

PR-to-objective association and the branch-pr workflow require a GitHub repo. Repo-less objectives use the direct strategy instead.

GitHub webhook events

Push events, PR events, and check-run status are only received for linked repos. Repo-less projects don't receive or display them.

How agents connect to a repo-less project

The only practical difference is how the agent identifies the project at the start of a run. Instead of a repo slug, pass the project UUID from the project settings page:

# Engineering project — linked to GitHub

planwright_set_repo(githubRepo: "your-org/your-repo")

# Any other project — no repo required

planwright_set_repo(projectId: "your-project-uuid")

Everything after that call — listing objectives, claiming work, posting plans, requesting acceptance — is identical. The MCP server doesn't distinguish between project types once the session is established.

The bigger picture: When every team in an organization — engineering, product, GTM, ops, compliance — runs objectives through the same planning model, Planwright becomes the strategic operating system for AI-native work, not just the engineering workflow. The audit chain, the acceptance gate, and the Strategic Focus taxonomy apply uniformly. One board, one evidence trail, one model of accountability.