The first version of your company's harness, in the repository: CLAUDE.md, skills, hooks, permissions and spec, plan and task templates. It is the most important deliverable of the engagement.
Software Engineering with Claude Code in 4 Weeks
A dedicated Base/labs architect for 4 weeks takes your team to build software with AI using a clear method. You get your first harness and a usage tutorial.
4 weeks full-time, or spread over up to 12 weeks · 1 dedicated Base/labs architect · 1 client team of up to 10 people
-
CLAUDE.mdproject context W1 -
.claude/skills/team procedures W2 -
.claude/agents/review subagents W2 -
.claude/settings.jsonpermissions and hooks W1 -
docs/specs/<feature>/spec.mddocument 1 W1 -
docs/specs/<feature>/plan.mddocument 2 W2 -
docs/specs/<feature>/tasks.mddocument 3 W2 -
docs/harness-tutorial.mdusage tutorial W4
- weeks
- 4
- dedicated architect
- 1
- people on the team
- up to 10
- review gates
- 4
Base/labs is Certified in the Claude Partner Network. Our team is Claude certified.
Companies with software engineering teams that want to use Claude Code as a way of working, not just a smarter autocomplete, starting with one team of up to 10 people. The goal is to ship real changes faster, without losing control of quality and security. The harness is the set of instructions, rules and checks that guides the AI agent on your company's code.
Claude Code without a method becomes "vibe coding": a vague request, the model guesses the requirements nobody stated, and the result is scope that grows mid-task, half-finished implementations and code that ignores the project's patterns. The bigger the change, the bigger the damage. The answer is not a better prompt but process and environment: first decide what to build, then how, and only then write code, inside a harness that limits and verifies what the agent does.
A senior Base/labs architect, named in the proposal, is dedicated to your team for the whole period. In case of a prolonged absence, a replacement of the same level steps in, with a handover.
Without a method
- A vague request, and the model guesses what nobody said
- Scope grows mid-task
- Half-finished implementation
- Code that ignores the project's patterns
With spec, plan and tasks
- The approved spec defines what to build
- The approved plan defines in which steps
- Each task is born with the test that validates it
- An independent reviewer checks against the spec
The spec-driven process, in practice
Instead of a loose prompt, every change goes through three documents (spec, plan and task list) and four review gates: spec approval, plan approval, review of each task and a final review before merge. The documents are files in the repository: they survive between sessions, go into the pull request and can be read by anyone on the team.
For each task
- Failing test
- Passing code
- Refactoring
- Subagent review
Contradiction with the spec? Work goes back to the spec.
1. Spec: what and why
Document 1. The agent runs design questions (shape, data sources, edge cases, what stays out of scope) and writes the specification. Gate 1: someone on the team reviews and approves the spec before any plan exists.
2. Plan: in which steps
Document 2. From the approved spec, the plan defines what "done" means and which files change. Gate 2: human review approves the plan before any code.
3. Tasks and code: one at a time
Document 3: the ordered task list, each task with the test that validates it. For each task: a failing test, the code that makes it pass, refactoring and, at gate 3, a review by a subagent that reads the change without the context of whoever wrote it. If a task contradicts the spec, work stops and goes back to the spec.
4. Final review: against the spec
Gate 4: an independent reviewer checks the whole change against the spec and the plan before merge, and a person approves.
The harness: the environment that controls the agent
The quality of what the agent delivers depends on the environment it works in, not just the model. Harness engineering is designing that environment. The most important deliverable of the 4 weeks is the first version of your company's harness, with a short tutorial on how to use it. The architect builds it together with the pilot team, and it lives in the repository.
Context
What the agent needs to know about the project: architecture, code standards, build and test commands and decisions already made, in a lean CLAUDE.md.
Memory
What must survive between sessions: specs, plans and conventions versioned in the repository, instead of depending on a conversation history.
Skills
Repeatable team procedures (opening a spec, reviewing a PR, generating a report), with deterministic steps where the result cannot vary.
Tools
What the agent can use: approved MCP servers, integration with CI, the task tracker and the repository, each with minimum permissions.
Constraints
Guardrails and a permissions policy that define what the agent does not do, with hooks for rules that admit no exceptions and clear autonomy limits.
Verification
Tests, automated checks and independent review that prove the change does what the spec asks before it reaches merge.
What we deliver
A short tutorial on how to use the harness day to day, for the team and for whoever joins later.
A Base/labs architect dedicated to the engagement for the whole period, leading it hands-on.
A before/after report, with cycle time, review time and rework measured from Git history. Four weeks is a short period for statistical conclusions, and the report says so.
In real use at the end of the 4 weeks: task and spec, implementation (feature, fix and hotfix), PR review, the CI/CD the team already has, QA validation and bug, vulnerability and performance detection with AI support.
Introduced and demonstrated, for the team to evolve later: long-running agents, MCP integrations and monitoring.
How it works
Before we start (outside the 4 weeks)
A few days of preparation before the start, with engineering leadership, security, identity and network teams and the repository owner. We settle the decisions that must exist before anyone installs the tool, choose the team (up to 10 people, including any QA, PO and tech lead who take part), the repository and the real features, and capture the baseline from Git history. The 4 weeks begin with everything ready.
Week 1: harness v1 and first spec
The architect runs a repository diagnosis and builds the first version of the harness. The team ends the week with the approved spec of a real feature, written with the agent.
Week 2: from spec to PR
Plan and tasks with tests, implementation with TDD, skills with deterministic steps and subagent review, pairing with the architect. Features, fixes and hotfixes go through the normal pull request flow, and the first changes reach merge. The harness is adjusted with every lesson learned.
Week 3: from PR to production
In real use: the team's CI/CD, QA validation, bug, vulnerability and performance detection on the pilot PRs and a second cycle of bigger changes that touch many files. Introduced and demonstrated: long-running agents, MCP integrations and monitoring. We start writing the tutorial from real usage.
Week 4: autonomy and handover
The team runs the full cycle on its own and the architect only reviews. Together we define the agents' autonomy limits. Your team takes over the harness and receives the final tutorial and the before/after report.
Frequently asked questions
The engagement
How does the architect's dedication work?
A Base/labs architect is dedicated to your team. The reference format is 4 weeks full-time. The 4 weeks can also be spread over up to 12 weeks: the architect's dedication is proportional, and each stage of the schedule stretches by the same proportion. We agree on that when designing the proposal.
What is the harness?
It is the environment around the coding agent: project context, memory between sessions, skills, tools, constraints and verification. Instead of relying on a perfect prompt, the team designs that environment once and reuses it on every task. The first version of your company's harness, with a short usage tutorial, is the main deliverable of the engagement.
Why a single team of up to 10 people?
Because a dedicated architect can closely follow a team of that size, on real work, during the 4 weeks. The harness left behind is the basis for taking the process to other teams afterwards.
Who counts among the 10 people?
Everyone who takes part in the work day to day: developers, QA, PO and tech lead.
Is this just training?
No. It is an engineering engagement: the team delivers real changes from your backlog, through the normal pull request flow, with the architect alongside.
How is this different from Claude Code's plan mode?
Plan mode produces a plan inside the session itself. In the spec-driven process, spec and plan are files in the repository, reviewed by people and reused across sessions and teams.
Do you use Spec Kit, Superpowers or BMAD?
We choose the flow based on your context. Open frameworks like these help structure the process, and what matters is that spec and plan live in the repository and go through review. For small changes, like a style tweak, the full process is overkill, and we teach the team to tell the difference.
Commercial
Are Claude licenses included? Who pays for usage?
No. Licenses and token consumption stay with your company, contracted directly with Anthropic or through your cloud marketplace. The required plan and the usage estimate are defined together at the start of the work.
See also: Claude: Governance and Cost
Is there an acceptance criterion or guarantee? What if the chosen feature is not finished?
We do not promise a productivity result. Acceptance is based on delivery of the agreed items: the harness, the tutorial and the before/after report. If the pilot feature does not fit in the 4 weeks, we adjust its scope, and that does not prevent acceptance.
Effort and logistics
How much of my team's time is committed? Does the team stop other deliveries?
The team carries on with its deliveries. The pilot features come from the current backlog, so they are not extra work. The additional time is for pairing with the architect and for reviews.
Is the work remote or on-site?
Remote by default. If it makes sense, on-site work can be agreed in the proposal.
How long between signing and starting?
To be agreed in the proposal, because it depends on access and on the initial platform, security, identity and network decisions, which are part of the preparation before the 4 weeks.
Technical
Does it work with my stack, a monorepo or a legacy system?
Claude Code works on Git repositories, from the terminal, and the method does not depend on a specific language. Monorepos and your team's repository and CI platforms are assessed during preparation. For legacy systems that need to be migrated, see the modernization offering.
See also: Legacy Modernization with Claude Code
What if the repository has little test coverage?
The diagnosis measures coverage. Where it is low, the tests that protect current behavior are written before the change, and we choose the pilot scope to fit the 4 weeks. If the repository calls for much more work than that, we point to the modernization offering.
See also: Legacy Modernization with Claude Code
Why Claude Code, and can the team keep using other tools?
The offering is built on Claude Code. The method (spec, plan, tasks and review) and the harness live in the repository, and your team can keep using other tools day to day.
Who maintains the harness when Claude Code changes?
Your team, which receives the harness and the tutorial at handover. If you want support with adjustments, the company can hire more hours of the architect.
Security and legal
Is my code used to train the model? What about LGPD?
According to Anthropic's public policies, inputs and outputs of commercial products (such as Claude Enterprise, the API and Claude Code under commercial terms) are not used to train its models by default, unless the customer opts in. What applies to your company are the terms you contract, and the model access route (Claude Enterprise, Console or a cloud such as Bedrock, Google Cloud or Foundry) is decided during preparation. The LGPD assessment is done with your legal and privacy team at the same stage.
What access does the architect need? Do you sign an NDA?
Minimum access to the pilot team's repository and CI, with no access to production, according to your security team's policy. Base/labs signs an NDA before the start.
Who owns the harness, skills and templates?
Your company. Everything produced in the engagement is versioned in your repository.
How do we stop the agent from doing something destructive?
The permissions policy defines what the agent can access and run. Sensitive files and destructive commands go on a deny list, and the rules that admit no exceptions become hooks, automated checks, instead of a request in the prompt. Every change goes through a pull request with human review.
Results
What productivity gain can I expect?
We do not promise a number. We measure before and after from Git history (cycle time, review time and rework) and deliver a report. Four weeks is a short period for statistical conclusions, and the report makes that clear.
After the project
How do I take the process to other teams?
The harness and the tutorial stay with your team and are the basis for the others. To support that expansion, the company can hire more hours of the architect or talk about the Certified Claude Squad.
See also: Certified Claude Squad
Can I hire more hours of the architect after the engagement ends?
Yes. There is no support included after the handover, but the company can hire more hours of the architect, on an ad hoc basis, for follow-up, for example to review the harness, answer questions or support the arrival of other teams.
Want to talk about this offering?
We start with a conversation to understand your scope. Each offering has a closed scope and price per phase.
