Rolling out Claude Code?

Claude Code loose on your codebase? We set it up properly.

Claude Code is a capable terminal agent and a sharp knife at the same time. A single engineer running it inside a monorepo can land a feature in an afternoon, or quietly commit a secret, drop a production table, and open a pull request that nobody reviewed. We set it up with the right permissions, the right MCP servers, the right pre-commit guardrails, and the audit trail an enterprise buyer actually needs.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

claude-code.log

still failing

What it is

Claude Code at a glance

Claude Code is Anthropic's agentic command-line tool for software engineering, released in 2024 and now broadly used inside and outside Anthropic. It runs in a terminal, reads your repository, edits files, runs tests, drives git and shell commands, and uses Claude 4.x models to reason about what to do next. It can be extended with Model Context Protocol servers that give it access to databases, issue trackers, cloud providers and internal APIs.

It ships for individual developers on the Anthropic API, through a Claude for Enterprise plan, and against Amazon Bedrock and Google Vertex AI for teams that need their own cloud account. Permissions, approved tools, MCP servers and audit logging are configured through settings.json files at user, project and enterprise levels. The combination of agent autonomy and shell access makes guardrails more than a nice-to-have.

Type
Agentic coding CLI with MCP server extensibility
First released
2024
License
Proprietary CLI, open-source SDK and MCP reference servers
Maintained by
Anthropic, PBC
Typically used for
Feature work, refactors, codebase exploration, bug triage, migration scripts, writing tests, driving CI/CD from a terminal, running internal agents over private data

The honest version

Claude Code pros and cons

No technology is all upside. This is where Claude Code earns its place and where it tends to cause trouble.

Where Claude Code is strong

  • Thinks in whole repositoriesClaude Code reads files, follows imports, runs greps and inspects the git history as part of answering a question. For non-trivial work that touches several files, the context quality is noticeably higher than an in-editor assistant.
  • MCP makes the agent useful against real systemsModel Context Protocol servers connect the agent to Postgres, GitHub, Jira, Slack, Sentry, your own HTTP APIs and more. The agent can propose changes based on real data rather than guessing.
  • Permission model you can actually configureAllowed tools, hook scripts, approval prompts and settings.json at user and enterprise levels mean you can give Claude Code real power without handing it the whole machine.
  • Bedrock and Vertex deployment optionsTeams that need data to stay inside an AWS or Google Cloud account can run Claude Code against Bedrock or Vertex, with the same CLI and the same enterprise controls.
  • Great for research and migrationMulti-file refactors, framework upgrades and codebase-wide edits are where the agent shows its edge, because it can plan, apply, test and iterate without a human driving every step.
  • Subagents and plan modeDelegating research or exploration to subagents, and running in plan mode before making changes, keeps longer sessions auditable and easier to supervise.

Where Claude Code bites

  • Agent autonomy plus shell access is a sharp edgeClaude Code can commit code, open pull requests and run arbitrary shell commands. The same capabilities that make it useful can wipe files or push secrets if permissions are left wide open.
  • MCP configuration is fiddlystdio transports, environment variables, path quoting and token scopes all have to be right. The error messages when something is wrong usually point at the symptom, not the cause.
  • Costs climb across a whole engineering orgEach engineer can burn through hundreds of thousands of tokens a week on normal work. The invoice is predictable once measured, and surprising the first time somebody looks.
  • Context budget is finiteA 200k token window sounds huge until the agent is reading a monolith. Auto-compaction keeps things moving but can lose detail that mattered, and long sessions are not always reliable.
  • Enterprise posture is a projectApproved tools, hook scripts, audit logging, SSO, data residency and Bedrock or Vertex wiring are all things to set up explicitly. The CLI ships with sensible defaults, not a finished governance story.
  • Guardrails on untrusted content are your jobPrompt injection in a README, a Jira ticket or a file in the repository can attempt to redirect the agent. Treating any read input as potentially hostile is the safe default.

Sound familiar?

Common Claude Code problems we fix

The Claude Code problems that bring people to us, and what you have once each one is fixed.

AWS_SECRET_ACCESS_KEY committed

Agent pushed secrets to a branch

A permissive permission policy let the agent read, stage and commit a .env file, and the commit landed before anyone reviewed the diff. Rotation and history rewrite now follow.

Secrets rotated, policy tightened

MCP server 'postgres' failed to start

MCP servers refuse to load on boot

A path that works on one machine is missing on another, or an environment variable needed by the server is only present in the shell after the CLI launches.

MCP boots cleanly on every machine

bedrock InvokeModel AccessDenied

Enterprise Bedrock access is not entitled

The account is signed up for Bedrock but not entitled for the specific Claude model, or the IAM role granting the CLI access lacks the bedrock:InvokeModel permission.

Entitlements and IAM aligned

$8,420/month, 14 seats

Spend has outrun the team's expectations

Agent sessions running longer than they need to, engineers choosing Opus for small tasks, and no cost dashboard mean the monthly invoice doubled before anyone noticed.

Model routing, budgets and dashboards in place

Bash(*) in allowedTools

Permission policy is effectively unlimited

A default settings.json left shell access unrestricted on every engineer's laptop, so the agent can rm, curl and push to whatever branch it pleases.

Tools scoped, approvals in place

context auto-compacting at 187k

Long sessions lose important context

Auto-compaction drops older messages to keep the conversation within the window, and sometimes drops the specific instruction that was holding the session on track.

Sessions split or pinned appropriately

PR opened with no review gate

Agent-opened pull requests merge themselves

Branch protection does not require a reviewer, the developer who prompted the agent also approves the PR, and AI-written code ships without a second pair of eyes.

Review required, humans in the loop

prompt injection via README

Untrusted content hijacks the agent

A dependency README or a Jira ticket contains instructions that override the system prompt, and the agent follows them unless the policy is explicit.

Input scoping and policy explicit

What we do

Claude Code help, from rollout to upkeep

Call us for one failed MCP setup, or let us roll Claude Code out across the whole engineering team.

Rescue
An agent that committed secrets, rewrote files it should not have, or opened a pull request nobody asked for. We stop the bleeding, rotate what has to be rotated, and leave the repo clean.
Enterprise rollout
SSO, settings.json at each level, allowed tools, hook scripts, audit logging, Bedrock or Vertex wiring, and training for the first wave of engineers.
MCP server setup
Reference MCP servers, your own custom MCPs for internal APIs, and the permission scopes to make them safe to leave running against production systems.
Guardrails
Pre-commit hooks, secret scanning, branch protection, approval policies and review gates so the agent speeds the team up without putting production at risk.
Cost control
Model routing, budget alerts, team-level usage dashboards and contract review, with the before and after spend measured against a real month of usage.
Managed Tech
Ongoing support across new Claude Code versions, MCP server updates, and the operational questions that come up as the agent gets deeper into daily workflows.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running Claude Code alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

Claude Code questions

Something else? Call+1 737 727 7327

Can you roll Claude Code out safely across our engineering team?

Yes. We start with a settings.json policy that disables the obviously dangerous tools by default, wire in SSO and audit logging, configure the MCP servers your team actually needs, and run a short enablement session with each squad. Rollout is typically a week or two for a medium organisation.

Our agent committed secrets to GitHub. What now?

First, rotate everything that was exposed, including service accounts that may have been used with those credentials. Then rewrite history to scrub the keys, force-push carefully, and run a secret scan across the whole repo. We handle the rotation and scrub, and tighten the permission policy so the same commit cannot happen again.

Why do my MCP servers fail to start?

Almost always a path, environment variable or token scope issue. The stdio handshake is strict about JSON over a clean pipe, so anything printing to stderr during startup, or an env var that is set in your shell but not where the CLI runs, will prevent the server from registering. We walk the logs and get each server booting reliably.

Should we run Claude Code via Bedrock, Vertex, or the Anthropic API directly?

If your compliance or data residency requirements point at an AWS or Google Cloud account, Bedrock or Vertex is the right choice and the CLI supports both. For teams without that constraint, the direct API is the simplest path and tends to get new features soonest. We help you pick based on your actual requirements rather than the loudest ones.

How do we stop prompt injection in Claude Code?

Treat any content the agent reads, including READMEs, issue descriptions, test fixtures and scraped pages, as untrusted. We configure tool permissions so the agent cannot take destructive actions without approval, scope MCP servers to least privilege, and add hooks that reject commands matching known dangerous patterns.

How much does Claude Code help cost?

A written quote is free. Tell us what the team is trying to run and what has gone sideways, and a price lands in your inbox before any configuration is touched.

Free estimate

Claude Code loose in your repo?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.