Independent browser-local gate · Source checked August 27, 2026
Is Code Context ready for a bounded rollout?
Review rollout, repository scope, indexing rights, data location, permissions, third-party agents, retrieval quality, change control, and deletion without connecting to Atlassian or your SCM.
Local result
Complete the evidence review
Eight gates remain. Begin with lifecycle, SCM scope, secrets, indexing rights, and data location before enabling code retrieval.
A passing checklist does not certify security or approve source transfer, agent actions, code changes, deployment, or merge. Start with read-only questions in low-risk repositories.
Continue by decision
Lifecycle and accessCheck open-beta rollout, editions, admin opt-in, search modes, and pilot sequence. GitHub vs. BitbucketCompare connectors, mirroring, indexing, permissions, freshness, and disablement. Coding-agent CLIBound third-party providers, source transfer, costs, tools, actions, and benchmarks. Security and indexingReview branches, tags, US storage, embeddings, permissions, limitations, and deletion.
Atlassian Code Context questions
No. It runs locally in the browser and does not inspect an Atlassian site, enable Code Context, connect an SCM, index code, run TWG CLI, send source to an agent, or change permissions.
No. It only shows that the eight planning gates were checked. Security, privacy, legal, licensing, procurement, records, platform, repository-owner, and change-management approvals remain separate.
Atlassian says it respects existing SCM permissions and validates them at query time. Verify those permissions and denial cases in your own GitHub or Bitbucket configuration.
No. Retrieval, file writes, tool calls, commands, network access, pull requests, deployments, and merge approval are separate authority boundaries.
No. Atlassian documents semantic search as discovery rather than a security or compliance scanner and notes that results have no current confidence signal.
Official references: launch, product and data details, setup, and benchmark.