Skip to content

remediate — AgentTeamsModule

Auto-correction support via standalone Copilot CLI.

Provides an optional remediation pass that runs after post-audit finds issues in generated agent team files. Uses the standalone copilot CLI in non-interactive mode, scopes file access to the generated team directory, then returns control to the main pipeline for a verification rerun.

Source: agentteams/remediate.py

Classes

RemediationResult

Source: agentteams/remediate.py

Outcome of a Copilot CLI remediation attempt.

Attributes:

  • attempted (bool) — True if the Copilot CLI was invoked.
  • succeeded (bool) — True if the CLI exited with code 0.
  • message (str) — Human-readable status message.
  • command (list[str]) — The exact command that was run.
  • stdout (str) — Standard output captured from the CLI.
  • stderr (str) — Standard error captured from the CLI.

Functions

run_copilot_autocorrect(*, output_dir, manifest, audit_result)

Source: agentteams/remediate.py

Invoke the standalone Copilot CLI to repair generated team files.

Args:

  • output_dir (Path, keyword-only) — Root directory containing generated team files.
  • manifest (dict[str, Any], keyword-only) — Team manifest from analyze.build_manifest().
  • audit_result (keyword-only) — AuditResult-like object containing findings.

Returns: RemediationResult — If the Copilot CLI is absent or exits non-zero, returns a result with succeeded=False; no exception is raised for those cases.

Raises: OSError — Only for unexpected OS-level subprocess failures (e.g. permission errors on the executable path). A missing Copilot CLI returns succeeded=False, not OSError.

Source: agentteams/remediate.py

Print a human-readable remediation summary to stdout.

Args:

  • result (RemediationResult) — Result from run_copilot_autocorrect().