diff --git a/docs/src/content/docs/reference/glossary.md b/docs/src/content/docs/reference/glossary.md
index 5e816c11806..0697a9d99b3 100644
--- a/docs/src/content/docs/reference/glossary.md
+++ b/docs/src/content/docs/reference/glossary.md
@@ -75,6 +75,10 @@ A service that implements the Model Context Protocol to provide specific capabil
 
 An MCP server field that controls startup criticality. By default every configured MCP server is startup-critical: the agent job fails if the server cannot be reached during the pre-flight connectivity check. Setting `required: false` marks a server as best-effort, so an unreachable server logs a warning and the workflow continues without it, rather than aborting the run. At least one server must still connect successfully for startup to proceed. See [MCPs Guide](/gh-aw/guides/mcps/).
 
+### Partial Results (MCP Logs)
+
+A response pattern used by the `gh aw` MCP server's `logs` tool when a gateway timeout or token budget guardrail prevents returning a complete result set. Instead of failing the call, the tool returns the JSON data it collected so far with `partial: true` and continuation parameters (such as `after_run_id`/`before_run_id`) so the caller can request the remaining results in a follow-up call. See [`gh aw` as an MCP Server](/gh-aw/reference/gh-aw-as-mcp-server/).
+
 ### QMD Documentation Search (`qmd:`)
 
 A built-in tool that provides vector similarity search over documentation files. Configured via `tools.qmd:` in frontmatter, the `qmd` tool runs [tobi/qmd](https://github.com/tobi/qmd) as an MCP server so agents can find relevant documentation by natural language query. The search index is built in a dedicated indexing job (which has `contents: read`) and shared with the agent job via `actions/cache`, so the agent job does not need `contents: read`. Supports indexing from repository checkouts, GitHub code search queries, and cache-only read-only mode. See [QMD Documentation Search](/gh-aw/reference/qmd/).
@@ -623,6 +627,14 @@ Sample CLI engine definitions bundled as reference patterns rather than official
 
 See [AI Engines Reference](/gh-aw/reference/engines/#pinning-a-specific-engine-version).
 
+### Post-Result Watchdog (`engine.harness.watchdog-timeout`)
+
+A harness setting that terminates a quiet child process after the agent has already produced a terminal safe output, such as `noop` or an ordinary task output (comment, label, push, pull request creation). Diagnostic safe outputs like `missing_tool`, `missing_data`, and `report_incomplete` are not terminal and do not arm the watchdog. The watchdog stays dormant until armed, and once armed, any stdout or stderr activity resets its inactivity clock — so a quiet process doing useful work can still be terminated, with the harness treating that termination as successful because the requested safe output already exists. Configure with `engine.harness.watchdog-timeout` (seconds) or the raw `GH_AW_HARNESS_WATCHDOG_TIMEOUT_MS` environment variable (milliseconds, default `120000`, min `50`, max `600000`). See [Harness Retry and Post-result Watchdog Policy](/gh-aw/reference/engines/#harness-retry-and-post-result-watchdog-policy) and [Environment Variables Reference](/gh-aw/reference/environment-variables/#shared-post-result-watchdog).
+
+### Harness Retry Runner
+
+A shared retry module used by the built-in Copilot, Claude, and Codex engine harnesses to re-launch the agent CLI with exponential backoff after a failed or interrupted run. Configurable via `engine.harness` fields (or `GH_AW_HARNESS_*` environment variables) for retry count and backoff multiplier/max-delay. Extracted as a common component so all harnesses share the same retry behavior instead of duplicating it per engine. See [AI Engines Reference](/gh-aw/reference/engines/#harness-retry-and-post-result-watchdog-policy).
+
 ### Anthropic Workload Identity Federation (WIF)
 
 A keyless authentication method for the Claude engine that uses short-lived GitHub OIDC tokens instead of a long-lived `ANTHROPIC_API_KEY` secret. Configured via [`engine.auth`](#engine-auth-engineauth) with `type: github-oidc` and `provider: anthropic`, along with Anthropic-specific IDs (`federation-rule-id`, `organization-id`, `service-account-id`, `workspace-id`) obtained from the Anthropic Console. Requires `permissions: id-token: write`. Available since v0.79.6. See [Authentication Reference](/gh-aw/reference/auth/#anthropic-workload-identity-federation-wif).
@@ -1129,6 +1141,10 @@ A failure category reported when one or more frontmatter skills could not be ins
 
 A GitHub Personal Access Token with granular permission control, specifying exactly which repositories the token can access and what permissions it has. Created at github.com/settings/personal-access-tokens.
 
+### Copilot-Only Artifacts (`--no-mcp`, `--no-agent`)
+
+Configuration files that `gh aw init` creates only when the Copilot engine is selected (`--engine copilot`): the Agentic Workflows custom agent (`.github/agents/agentic-workflows.md`) and MCP server integration. Both are enabled by default with the Copilot engine; pass `--no-mcp` to skip MCP server integration or `--no-agent` to skip custom agent creation. Non-Copilot engines skip these artifacts automatically. See [CLI Reference](/gh-aw/setup/cli/#init).
+
 ### `RUNNER_TEMP` / `${{ runner.temp }}`
 
 A GitHub Actions environment variable pointing to a per-job temporary directory on the runner. Agentic workflows store compiled scripts and runtime artifacts under `${RUNNER_TEMP}/gh-aw/` for compatibility with self-hosted runners that may not have write access to system directories. In shell `run:` blocks, use the shell variable form `${RUNNER_TEMP}`; in `with:` or `env:` YAML fields, use the expression form `${{ runner.temp }}`.