✨ feat: support timeline --before/--after filters

README.md

@@ -64,15 +64,19 @@ npx skills add https://github.com/ShigureLab/gh-llm --skill github-conversation
 Read a PR's full timeline — metadata, comments, reviews, checks — with progressive expansion:
 
 ```bash
-# Show first + last timeline pages with actionable hints
+# Initial read: show first + last timeline pages with actionable hints
 gh-llm pr view 77900 --repo PaddlePaddle/Paddle
 gh llm pr view 77900 --repo PaddlePaddle/Paddle
 
+# Later incremental read: reuse the previous frontmatter `fetched_at`
+gh-llm pr view 77900 --repo PaddlePaddle/Paddle --after 2026-04-08T02:41:17Z
+
 # Show selected regions only
 gh-llm pr view 77900 --repo PaddlePaddle/Paddle --show timeline,checks
 
 # Expand one hidden timeline page
 gh-llm pr timeline-expand 2 --pr 77900 --repo PaddlePaddle/Paddle
+gh-llm pr timeline-expand 2 --pr 77900 --repo PaddlePaddle/Paddle --after 2026-04-08T02:41:17Z
 
 # Auto-expand folded content in default/timeline view
 gh-llm pr view 77900 --repo PaddlePaddle/Paddle --expand resolved,minimized
@@ -118,12 +122,16 @@ Issue reading works the same way as PR reading — timeline view with progressiv
 
 ```bash
 gh-llm issue view 77924 --repo PaddlePaddle/Paddle
+gh-llm issue view 77924 --repo PaddlePaddle/Paddle --after 2026-04-08T02:41:17Z
 gh-llm issue timeline-expand 2 --issue 77924 --repo PaddlePaddle/Paddle
+gh-llm issue timeline-expand 2 --issue 77924 --repo PaddlePaddle/Paddle --after 2026-04-08T02:41:17Z
 gh-llm issue comment-expand IC_xxx --issue 77924 --repo PaddlePaddle/Paddle
 gh-llm issue view 77924 --repo PaddlePaddle/Paddle --expand minimized,details
 gh-llm issue view 77924 --repo PaddlePaddle/Paddle --show meta,description
 ```
 
+For incremental follow-ups, copy the previous output's `fetched_at` value into `--after <fetched_at>`. `--before` is also available when you want to inspect only older timeline slices.
+
 When `--show` does not include `timeline` (for example `--show meta`, `--show summary`, or `--show actions`), both `pr view` and `issue view` stay on the lightweight metadata path and skip timeline bootstrap.
 
 Use `--show` to choose which output sections to render. Use `--expand` to automatically open folded content within those sections.
@@ -309,6 +317,8 @@ This supports the normal flow where one review contains multiple inline comments
 All output follows consistent formatting rules so both humans and LLMs can parse it reliably:
 
 - **Metadata** is rendered as YAML-style frontmatter at the top of PR/issue views.
+- Frontmatter includes `fetched_at`, so the next incremental read can use `--after <fetched_at>`.
+- When timeline filtering is active, frontmatter also includes `timeline_after` / `timeline_before` and filtered vs unfiltered event counts.
 - **Description** is wrapped in `<description>...</description>` tags.
 - **Comment bodies** use `<comment>...</comment>` tags to avoid markdown fence ambiguity with code blocks inside comments.
 - **Hidden timeline sections** are separated by `---` dividers and include ready-to-run expand commands to load the omitted content.

skills/github-conversation/SKILL.md

@@ -109,11 +109,15 @@ Use this before forking or opening a PR when you need the default branch, onboar
 
 ```bash
 gh-llm pr view <pr> --repo <owner/repo>
+gh-llm pr view <pr> --repo <owner/repo> --after <previous_fetched_at>
 gh-llm pr timeline-expand <page> --pr <pr> --repo <owner/repo>
+gh-llm pr timeline-expand <page> --pr <pr> --repo <owner/repo> --after <previous_fetched_at>
 gh-llm pr review-expand <PRR_id[,PRR_id...]> --pr <pr> --repo <owner/repo>
 gh-llm pr checks --pr <pr> --repo <owner/repo>
 ```
 
+Use plain `view` for the first pass. On follow-up reads, reuse the previous frontmatter `fetched_at` as `--after <previous_fetched_at>` for an incremental timeline refresh.
+
 ### Prepare a PR body
 
 ```bash
@@ -131,10 +135,13 @@ Use this before `gh pr create` when you need to load a repo PR template, append
 
 ```bash
 gh-llm issue view <issue> --repo <owner/repo>
+gh-llm issue view <issue> --repo <owner/repo> --after <previous_fetched_at>
 gh-llm issue timeline-expand <page> --issue <issue> --repo <owner/repo>
+gh-llm issue timeline-expand <page> --issue <issue> --repo <owner/repo> --after <previous_fetched_at>
 ```
 
 For lightweight inspection, prefer non-timeline `--show` combinations such as `--show meta`, `--show summary`, or `--show actions`; `gh-llm` keeps those paths on metadata-only loading unless `timeline` is explicitly requested.
+Frontmatter includes `fetched_at`, plus `timeline_after` / `timeline_before` and filtered vs unfiltered counts when timeline filtering is active.
 
 ### Write simple updates
 

src/gh_llm/commands/issue.py

@@ -5,7 +5,9 @@
 
 from gh_llm.commands.options import (
     add_body_input_arguments,
+    add_timeline_window_arguments,
     maybe_resolve_subject,
+    parse_timeline_window,
     raise_unknown_option_value,
     resolve_file_or_inline_text,
     resolve_subject,
@@ -65,6 +67,7 @@ def register_issue_parser(subparsers: Any) -> None:
         default=[],
         help="auto-expand folded content: minimized, details, all (comma-separated or repeatable)",
     )
+    add_timeline_window_arguments(view_parser)
     view_parser.set_defaults(handler=cmd_issue_view)
 
     timeline_expand_parser = issue_subparsers.add_parser("timeline-expand", help="load one timeline page by number")
@@ -78,6 +81,7 @@ def register_issue_parser(subparsers: Any) -> None:
         default=[],
         help="auto-expand folded content: minimized, details, all (comma-separated or repeatable)",
     )
+    add_timeline_window_arguments(timeline_expand_parser)
     timeline_expand_parser.set_defaults(handler=cmd_issue_timeline_expand)
 
     details_expand_parser = issue_subparsers.add_parser(
@@ -88,6 +92,7 @@ def register_issue_parser(subparsers: Any) -> None:
     details_expand_parser.add_argument("--issue", help="Issue number/url")
     details_expand_parser.add_argument("--repo", help="repository in OWNER/REPO format")
     details_expand_parser.add_argument("--page-size", type=int, help="timeline entries per page")
+    add_timeline_window_arguments(details_expand_parser)
     details_expand_parser.set_defaults(handler=cmd_issue_details_expand)
 
     comment_edit_parser = issue_subparsers.add_parser("comment-edit", help="edit one issue comment by node id")
@@ -113,6 +118,7 @@ def cmd_issue_view(args: Any) -> int:
     page_size = int(args.page_size)
     expand = _parse_expand_options(raw_values=list(getattr(args, "expand", [])))
     show = _parse_show_options(raw_values=list(getattr(args, "show", [])))
+    timeline_window = _resolve_timeline_window(args)
     client = GitHubClient()
     pager = TimelinePager(client)
 
@@ -126,6 +132,7 @@ def cmd_issue_view(args: Any) -> int:
         context, first_page, last_page = pager.build_initial(
             meta,
             page_size=page_size,
+            timeline_window=timeline_window,
             show_minimized_details=expand.minimized,
             show_details_blocks=expand.details,
         )
@@ -194,8 +201,14 @@ def print_block(lines: list[str]) -> None:
 def cmd_issue_timeline_expand(args: Any) -> int:
     client = GitHubClient()
     pager = TimelinePager(client)
-    context, meta = _resolve_context_and_meta(client=client, pager=pager, args=args)
     expand = _parse_expand_options(raw_values=list(getattr(args, "expand", [])))
+    context, meta = _resolve_context_and_meta(
+        client=client,
+        pager=pager,
+        args=args,
+        show_minimized_details=expand.minimized,
+        show_details_blocks=expand.details,
+    )
 
     page = pager.fetch_page(
         meta=meta,
@@ -217,13 +230,19 @@ def cmd_issue_timeline_expand(args: Any) -> int:
 def cmd_issue_details_expand(args: Any) -> int:
     client = GitHubClient()
     pager = TimelinePager(client)
-    context, meta = _resolve_context_and_meta(client=client, pager=pager, args=args)
+    context, meta = _resolve_context_and_meta(
+        client=client,
+        pager=pager,
+        args=args,
+        show_minimized_details=True,
+        show_details_blocks=True,
+    )
 
     index = int(args.index)
-    if index < 1 or index > context.total_count:
+    page_number = _resolve_timeline_page_for_index(context=context, index=index)
+    if page_number is None:
         raise RuntimeError(f"invalid event index {index}, expected in 1..{context.total_count}")
 
-    page_number = ((index - 1) // context.page_size) + 1
     page = pager.fetch_page(
         meta=meta,
         context=context,
@@ -232,10 +251,12 @@ def cmd_issue_details_expand(args: Any) -> int:
         show_details_blocks=True,
         diff_hunk_lines=None,
     )
-    page_start = (page_number - 1) * context.page_size + 1
-    offset = index - page_start
-    if offset < 0 or offset >= len(page.items):
-        raise RuntimeError("event index is outside loaded page range")
+    try:
+        offset = page.absolute_indexes.index(index)
+    except ValueError:
+        raise RuntimeError("event index is outside loaded page range") from None
+    except AttributeError as error:  # pragma: no cover - defensive fallback
+        raise RuntimeError("event index is outside loaded page range") from error
 
     for line in render_event_detail_blocks(index=index, event=page.items[offset]):
         print(line)
@@ -282,15 +303,41 @@ def _resolve_optional_issue(*, client: GitHubClient, args: Any) -> PullRequestMe
 
 
 def _resolve_context_and_meta(
-    *, client: GitHubClient, pager: TimelinePager, args: Any
+    *,
+    client: GitHubClient,
+    pager: TimelinePager,
+    args: Any,
+    show_minimized_details: bool = False,
+    show_details_blocks: bool = False,
 ) -> tuple[TimelineContext, PullRequestMeta]:
     page_size = getattr(args, "page_size", None)
     effective_page_size = DEFAULT_PAGE_SIZE if page_size is None else int(page_size)
     meta = _resolve_issue_meta(client=client, args=args)
-    context, _, _ = pager.build_initial(meta=meta, page_size=effective_page_size)
+    context, _, _ = pager.build_initial(
+        meta=meta,
+        page_size=effective_page_size,
+        timeline_window=_resolve_timeline_window(args),
+        show_minimized_details=show_minimized_details,
+        show_details_blocks=show_details_blocks,
+    )
     return context, meta
 
 
+def _resolve_timeline_window(args: Any):
+    return parse_timeline_window(after=getattr(args, "after", None), before=getattr(args, "before", None))
+
+
+def _resolve_timeline_page_for_index(*, context: TimelineContext, index: int) -> int | None:
+    if context.timeline_filtered:
+        for page_number, page in context.filtered_pages.items():
+            if index in page.absolute_indexes:
+                return page_number
+        return None
+    if index < 1 or index > context.total_count:
+        return None
+    return ((index - 1) // context.page_size) + 1
+
+
 def _parse_expand_options(*, raw_values: list[str]) -> _ExpandOptions:
     minimized = False
     details = False

src/gh_llm/commands/options.py

@@ -1,10 +1,13 @@
 from __future__ import annotations
 
 import sys
+from datetime import UTC, datetime
 from difflib import get_close_matches
 from pathlib import Path
 from typing import TYPE_CHECKING, Any, NoReturn
 
+from gh_llm.models import TimelineWindow
+
 if TYPE_CHECKING:
     from collections.abc import Callable
 
@@ -29,6 +32,40 @@ def add_body_input_arguments(
     )
 
 
+def add_timeline_window_arguments(parser: Any) -> None:
+    parser.add_argument(
+        "--after",
+        help="only include timeline events strictly after this ISO 8601 / RFC3339 timestamp",
+    )
+    parser.add_argument(
+        "--before",
+        help="only include timeline events strictly before this ISO 8601 / RFC3339 timestamp",
+    )
+
+
+def parse_timeline_window(*, after: str | None, before: str | None) -> TimelineWindow:
+    after_value = _parse_timestamp(raw=after, flag="--after")
+    before_value = _parse_timestamp(raw=before, flag="--before")
+    if after_value is not None and before_value is not None and after_value >= before_value:
+        raise RuntimeError("invalid time range: `--after` must be earlier than `--before`")
+    return TimelineWindow(
+        after=after_value,
+        before=before_value,
+        after_text=(format_timestamp_utc(after_value) if after_value is not None else None),
+        before_text=(format_timestamp_utc(before_value) if before_value is not None else None),
+    )
+
+
+def format_timestamp_utc(value: datetime) -> str:
+    utc_value = value.astimezone(UTC)
+    timespec = "microseconds" if utc_value.microsecond else "seconds"
+    return utc_value.isoformat(timespec=timespec).replace("+00:00", "Z")
+
+
+def current_timestamp_utc() -> str:
+    return format_timestamp_utc(datetime.now(UTC).replace(microsecond=0))
+
+
 def read_text_from_path_or_stdin(path: str) -> str:
     if path == "-":
         return sys.stdin.read()
@@ -94,3 +131,20 @@ def raise_unknown_option_value(
     suggest_text = f" Did you mean '{suggestion[0]}'?" if suggestion else ""
     valid_text = ", ".join(valid_values)
     raise RuntimeError(f"unknown {flag} option: {token}. Valid values: {valid_text}.{suggest_text}")
+
+
+def _parse_timestamp(*, raw: str | None, flag: str) -> datetime | None:
+    if raw is None:
+        return None
+    normalized = raw.strip()
+    if not normalized:
+        raise RuntimeError(f"{flag} requires a timestamp value")
+    if normalized.endswith(("Z", "z")):
+        normalized = normalized[:-1] + "+00:00"
+    try:
+        value = datetime.fromisoformat(normalized)
+    except ValueError as error:
+        raise RuntimeError(f"invalid {flag} timestamp: {raw}") from error
+    if value.tzinfo is None or value.utcoffset() is None:
+        raise RuntimeError(f"invalid {flag} timestamp: {raw}")
+    return value.astimezone(UTC)