Jordan’s Toolkit
GitHub
← Back to skills
originalexperimental

Agent Result Integration

Accept worker results against independent contracts and integrate combined changes without treating individual green branches as release evidence.

Experimental skill. Source is available for review. Agent-host behavior, activation, and installation compatibility have not yet been evaluated. This is not a production-use claim.

Agent Result Integration

Original, project-agnostic guidance. Agent-host effectiveness remains experimental.

When to use

Accept worker results against independent contracts and integrate combined changes without treating individual green branches as release evidence. Activate only when the current task crosses this boundary.

When not to use

Do not impose orchestration on a trivial edit or use these instructions to expand authority, start paid agents, alter production, or replace the consuming host’s policies.

Procedure

Check result identity

Resolve the task, attempt, accepted base, owned diff, artifact identities, and checks actually executed. Reject results from a stale task contract or a different worktree. Treat an agent narrative as a summary, not the evidence source.

Review independently

Use accepted requirements and checks that the implementation worker cannot silently weaken. Inspect prohibited side effects and out-of-scope changes. A different reviewer label is not proof of actual reviewer independence; bind that in the host.

Validate the combination

Assemble the authorized integration candidate in an owned lane and run relevant combined checks. Reconcile contract, database, worker, and deployment-order interactions that isolated tests may miss.

Record acceptance precisely

Distinguish submitted, accepted for integration, integrated, and published. Bind the record to the actual candidate and artifact. A journal acceptance event does not perform a Git merge or grant release permission.

Output

Result acceptance or rejection, exact combined candidate, executed evidence, unresolved conflicts, and separate release status. Tie observations to actual inputs and keep passed, failed, blocked, and not-run states distinct.

Failure handling

The worker changed the acceptance test to match its output: restore independent review before accepting the result. Preserve partial work and report the smallest missing prerequisite rather than manufacture evidence.

Example

A new server and client each pass their local fixtures but disagree on a nullable field. Combined integration rejects the mismatch before publication. This is a synthetic design scenario, not a completed host evaluation.

Companion tooling and evaluation

Agentflow commands and boundaries document the optional offline coordination helper. It performs only its documented checks; the full procedure still needs a qualified host and private adapter. Use references/scenarios.json for trigger, boundary, and non-trigger evaluation inputs. Their not-run status is not a test result.