Text of github.github.com › gh-aw › reference › artifacts as Styx crawled it on Nov 19, 2026. The page may have changed since. Only its words are kept, without pictures or layout, and only the first part of a long page.

Open the live page

Artifacts | GitHub Agentic Workflows

Skip to main content Artifacts GitHub Agentic Workflows upload several artifacts during workflow execution. This reference documents every artifact name, its contents, and how to access the data — especially for downstream workflows that use gh run download directly instead of gh aw logs . The agent and agent-output-fallback uploads require successful built-in secret redaction and every configured custom secret-masking step, regardless of whether the agent firewall is enabled.

Failed, skipped, cancelled, or missing redaction steps block both uploads, including custom steps with continue-on-error: true . An engine failure alone does not block these uploads when redaction succeeds. Cross-run artifact trust Section titled “Cross-run artifact trust” The built-in safe-output handoff consumes current-run structured outputs through configured handlers. A custom privileged workflow_run consumer downloading another run’s artifacts crosses a separate trust boundary.

Matching the repository ID, workflow name, branch, or successful conclusion does not prove that the producer ran trusted code: a base-repository workflow can process a fork PR. Redaction removes matching secrets; it does not authenticate artifact content or make it executable code. For downstream consumers: Validate the intended producer repository, workflow, event, run ID, and immutable source SHA against the consumer’s trust policy.

A producer-supplied manifest alone is not independent evidence of provenance. Download to an isolated staging directory, not over the checkout or a directory containing scripts/configuration used by subsequent steps. Reject unexpected paths, traversal, links, and files outside the expected layout. Parse only expected data formats with schema, size, and value limits. Do not source scripts, run binaries, import modules, or execute package-install/build hooks from an untrusted artifact.

Keep credentials and write permissions out of jobs that must execute untrusted output. Pass only validated data into a separate narrowly scoped privileged operation. Artifact names and checksums identify selected content, not whether that content is trustworthy. These checks are consumer responsibilities; generic download-and-execute steps are not made safe by compiling the surrounding workflow. Quick Reference Section titled “Quick Reference” Artifact Name Constant Type Description agent constants.AgentArtifactName Source: pkg/constants/job_constants.go Multi-file Unified agent job outputs (logs, safe outputs, token usage summary) agent-output-fallback constants.AgentOutputFallbackArtifactName Multi-file Small dedicated copy of processed agent output, raw safe-output NDJSON, execution evidence, token usage, and model-routing records, used when the larger agent upload fails or times out activation constants.ActivationArtifactName Multi-file Activation job output ( aw_info.json , prompt.txt , rate limits) firewall-audit-logs constants.FirewallAuditArtifactName Source: pkg/constants/constants.go Multi-file AWF firewall audit/observability logs (token usage, network policy, audit trail) detection constants.DetectionArtifactName Conditional Legacy inline engine ( features.gh-aw-detection: false ): single-file detection.log .

The default external gh-aw-detection engine: multi-file detection_result.json + step-summary.md ; detection.log is intentionally not uploaded (see below) safe-output constants.SafeOutputArtifactName Legacy/back-compat Historical standalone safe output artifact ( safe_output.jsonl ); in current compiled workflows this content is included in the unified agent artifact instead agent-output constants.AgentOutputArtifactName Legacy/back-compat Historical standalone agent output artifact ( agent_output.json ); in current compiled workflows this content is included in the unified agent artifact instead info constants.InfoArtifactName Single-file Standalone copy of the workflow run information ( aw_info.json ), uploaded by the activation job in addition to the copy bundled in the activation artifact aw-info — Legacy/back-compat Historical standalone engine-configuration artifact ( aw_info.json ); current compiled workflows upload this as info instead prompt — Legacy/back-compat Historical standalone prompt artifact ( prompt.txt ); in current compiled workflows prompt.txt is included in the activation artifact instead experiment constants.ExperimentArtifactName Multi-file A/B experiment state ( state.json ) uploaded by the activation job when experiments are declared in the frontmatter usage constants.UsageArtifactName Multi-file Compact conclusion-job artifact with workflow-run metadata and token-usage files used by lightweight reporting and forecasting paths graders constants.GradersArtifactName Conditional Combined built-in and custom JavaScript results and data-only manifest from the separate graders job evals constants.EvalsArtifactName Single-file BinEval evaluation results ( evals.jsonl ) uploaded by the evals job when evals are declared in the workflow frontmatter safe-outputs-items constants.SafeOutputItemsArtifactName Multi-file Safe output items manifest ( safe-output-items.jsonl ), temporary ID map ( temporary-id-map.json ), and failure diagnostics ( safe-output-errors.json , written only when the Process Safe Outputs step fails) code-scanning-sarif constants.SarifArtifactName Single-file SARIF file for code scanning results Legacy artifact names Section titled “Legacy artifact names” Artifact Sets Section titled “Artifact Sets” The gh aw logs and gh aw audit commands support --artifacts to download only specific artifact groups. logs defaults to usage ; audit defaults to all .

Set Name Artifacts Downloaded Use Case all Everything Full analysis agent agent , agent-output-fallback Agent logs and outputs activation activation Activation data ( aw_info.json , prompt.txt ) info info Compact workflow-run information ( aw_info.json ) firewall agent Network policy and firewall audit data mcp agent MCP gateway traffic logs detection detection Threat detection output experiment experiment , usage A/B experiment state (only present when experiments are declared) usage usage Compact conclusion-job artifact for lightweight reporting and forecasting evals usage BinEval evaluation results (only present when evals are declared) graders usage , graders , legacy agent artifacts Default built-ins and optional combined custom results github-api activation , agent GitHub API rate limit logs delivery safe-outputs-items , agent-output-fallback , usage Executed safe-output manifests, compact agent output, and conclusion evidence for delivery reporting work-queue activation , agent Queue snapshots, finish intents, and workflow logs for coordinator operation reporting Terminal window # Download agent and firewall artifacts only gh aw logs <run-id> --artifacts agent --artifacts firewall The former outcomes set was replaced by delivery without a compatibility alias.

Use gh aw logs <run-id> --delivery to report execution evidence, or --adoption to include later resource decisions. See Delivery and adoption for evidence and refresh semantics. firewall-audit-logs Section titled “firewall-audit-logs” The firewall-audit-logs artifact is uploaded by all firewall-enabled workflows . It contains AWF (Agent Workflow Firewall) structured audit and observability logs. When running AWF v0.28.59 or newer, the agent artifact also contains awf-resolved-config.json when AWF publishes AWF_RESOLVED_CONFIG .

AWF must publish a path matching /run/awf-config-audit/run-<six-alphanumeric>/awf-resolved-config.json ; the host-side redaction step validates the protected source’s ownership, mode, and regular-file link count, sanitizes sensitive fields, and stages only the sanitized copy at /tmp/gh-aw/awf-resolved-config.json before upload. Older AWF versions remain compatible and simply omit this file; omission is not reported as collected evidence.

Directory Structure Section titled “D

The rest of this page is on the live site.