iseewhatyoudid

iseewhatyoudid is a small reminder that your work adds up. When you are feeling as though you have not done enough, it lets you look back over the issues and pull requests you have contributed, step by step and over time.

It creates an interactive HTML dashboard of a GitHub user's activity: issues and pull requests they opened, status changes on those items, issue and pull request comments they wrote, and commit contributions. Activity is grouped by year, month, week, and day across every visible repository by default.

A gentle reminder, not a scorecard

This tool is intentionally a double-edged one. Seeing your past activity can help you recognise work that was easy to forget. It can also reinforce a sense of being unproductive when activity is low, when GitHub does not capture the work you did, or when you are simply going through a quieter period.

Take the dashboard with a grain of salt. GitHub activity does not measure a person's worth, pleasure, or happiness—and it is not a complete measure of their work or impact. Rest, learning, care, conversations, planning, and work away from GitHub all matter too.

Installation

pip install iseewhatyoudid

With uv:

uv tool install iseewhatyoudid

Python 3.10 or later is required.

Usage

By default, the dashboard shows all activity for the GitHub account authenticated in gh.

iseewhatyoudid dashboard

Pass --user to inspect a different GitHub user. Repository scopes are optional filters: an organization (--org) or an individual repository (--include-repo).

iseewhatyoudid dashboard --user octocat --include-repo octo-org/example

To inspect all repositories in an organization:

iseewhatyoudid dashboard --user octocat --org octo-org

The command writes iseewhatyoudid.html in the current directory. Open it in a browser to explore a personal history of how your work accumulated:

• a cumulative chart that only moves forward, with contribution milestones;
• a one-square-per-day activity tapestry;
• active-time, repository, conversation, and return-after-a-break insights;
• issue and pull-request lifecycle views;
• contribution variety, repository stewardship, and project exploration;
• personal records, first contributions, and yearly chapter cards;
• a long-term monthly rhythm; and
• links that help rediscover older issues and pull requests;
• commit-summary work types, including features, fixes, documentation, tests, refactoring, security, accessibility, CI, maintenance, and more;
• commit composition and cumulative work-type charts, a filterable commit heatmap, repository/type matrix, and project-role views;
• care-work summaries, milestones, maintenance seasons, recurring topics, and conventional-prefix coverage; and
• a searchable, linked commit-summary gallery plus locally derived personal insights.

The detailed stacked views for years, recent months, weeks, and days remain available near the end of the report.

Authentication

The command uses the GitHub CLI for authentication and GraphQL requests. Install gh and sign in once before running the dashboard. Its browser-based sign-in flow stores the credential in your system credential store when one is available.

gh auth login
iseewhatyoudid dashboard --org my-organization

When no --user is given, the account authenticated in gh is used. Make sure that account can access the repositories you include.

Scoping and filtering

--org, --include-repo, and --exclude-repo can each be repeated. Without an organization or included repository, all repositories are included. Otherwise, included repositories are combined with organization repositories; exclusions always take precedence.

# Activity in two organizations, plus one standalone repository.
iseewhatyoudid dashboard \
  --user octocat \
  --org octo-org \
  --org another-org \
  --include-repo someone/project \
  --exclude-repo octo-org/archived-project

Run iseewhatyoudid dashboard --help to see every option. While GitHub data is loading, the command displays a progress bar with fetched and total page counts. Logging is disabled by default. Add -v for Rich-formatted API logging, -vv for debug logging, or use --log-level to enable an explicit logging level. Add --refresh to bypass the short-lived cache when you need the very latest data.

Using local clones for deeper commit history

If your public repositories are already cloned below one or more directories, point the dashboard at them with --repos-dir:

iseewhatyoudid dashboard --repos-dir ~/dev

The option can be repeated. The command recursively discovers Git clones, normalizes each origin remote that points to github.com in the selected scope, and asks GitHub for only its canonical repository name, privacy status, and default-branch name. Other remotes—such as contributor forks added while reviewing pull requests—are ignored. Repositories do not need to appear in GitHub's contribution list to qualify. Discovery continues into the selected directory when that directory is itself a Git repository, so a Git-backed container of nested clones works as expected. The command then reads the local default branch without fetching, checking out, or changing the clone. Matched local histories replace commit-summary API requests; unmatched repositories keep the bounded GitHub fallback.

For the authenticated user, commits match the repository's configured Git email, GitHub's <login>@users.noreply.github.com forms, and mailmap aliases. Add old or additional identities explicitly when needed:

iseewhatyoudid dashboard \
  --repos-dir ~/work \
  --repos-dir ~/projects \
  --commit-author-email old-address@example.com

--commit-author-email can also be repeated. When --user selects another person, local Git configuration is deliberately not assumed to belong to that person; pass their known author emails explicitly. GitHub noreply addresses matching the selected login continue to work automatically.

HTML dashboard

Use --output-html to choose a different destination for the dashboard.

iseewhatyoudid dashboard --output-html activity.html

The report uses Chart.js from its CDN, so opening it requires network access. Hover over bars to inspect values or click legend entries to show and hide activity categories. Summary cards show totals and yearly/monthly averages, while a GitHub-style grid shows daily activity over the last year.

Cache

Activity is cached in the operating system's user cache directory, as resolved by platformdirs. After the initial history fetch, later runs request only the newest issue, pull request, and comment pages, refresh items that are still open or were closed recently, reuse completed commit years, and update the current year's commits. Merged pull requests and items closed for more than 30 days are treated as stable. Results fetched during the last 15 minutes are reused entirely, making quick dashboard reruns local-only after the authenticated username has been resolved.

The cache stores only the GitHub IDs, timestamps, states, repository names, titles, URLs, numbers, author logins, commit SHAs, and single-line commit summaries needed to rebuild the dashboard and its rediscovery links. It does not store comment bodies, issue descriptions, multi-line commit messages, patches, file contents, credentials, or local filesystem paths. For a matched local clone it additionally stores the repository's last inspected commit, reference name, shallow status, and a one-way fingerprint of the author identities so unchanged histories require no commit scan on the next run.

Commit-summary collection

Local clones are the preferred source when --repos-dir is used. Full clones provide up to the newest 2,000 matching commits from the locally available default-branch history per repository. Shallow clones provide only the history present on disk and are labeled as shallow in the dashboard. New descendants are scanned incrementally from the previously cached head. Rewritten histories and changed author identities are rescanned; unchanged shallow clones are reused, while deepened clones are detected and rescanned. Commit records remain cached by immutable SHA, and older records are trimmed when a repository exceeds the 2,000-commit limit.

For repositories without a usable local clone, commit-summary analysis is deliberately bounded so it does not turn dashboard generation into a remote repository crawl. Within the selected dashboard scope, the collector ranks public repositories using the contribution counts already available and samples at most 20 repositories. Repositories are batched ten at a time in GraphQL requests. The first import retrieves only the newest 100 default-branch commits attributed to the selected GitHub user in each sampled repository.

Commits are cached by immutable SHA. A normal refresh requests the newest page and stops as soon as it sees a cached SHA. If more than a full page of new commits appeared since the previous run, it can follow at most three pages per repository. The dashboard reports how many histories came from local clones, how many were shallow, and how many used the bounded GitHub sample.

The classifier recognizes forgiving Conventional Commit-style prefixes such as feat, feature, fix, docs, test, tests, refactor, perf, security, a11y, i18n, deps, ci, build, chore, and revert, with optional scopes such as feat(parser): .... Unrecognized summaries stay in a normal “Other” category; they are not treated as malformed.

Default merge subjects such as Merge branch ... and Merge pull request ... are excluded from commit-summary analysis so routine integration commits do not dominate work types or recurring themes. Custom merge commit subjects remain part of the analysis.

Private repositories are excluded from the cache and dashboard. GitHub's complete user-wide issue, pull-request, and comment connections do not support a public-only filter, so an authenticated response can transiently contain private nodes. The collector checks each repository's isPrivate flag and discards those nodes before adding anything to its history maps or writing a cache checkpoint. Private commit-contribution groups are discarded in the same way. Older cache schemas are invalidated and overwritten before collection starts.

Current limitations

Closed and merged dates describe status changes on issues and pull requests authored by the selected user; GitHub does not expose a global stream of items that the user personally closed or merged. Comments include issue comments and top-level pull-request conversation comments, but not inline review comments.

GitHub's contribution API returns commit activity from at most 100 repositories per year. For unusually active single repositories, the dashboard combines the first and last 100 commit-days returned by GitHub and warns when the API still truncates the year.

Commit-summary analysis covers only public default branches. Local clones can be stale until they are updated outside this program, and shallow clones omit history that is absent from disk. The GitHub fallback can omit older commits beyond its initial sample. Both sources omit unmerged branch commits and commits no longer reachable after rebases or force-pushes. Commits whose author email does not match a configured identity can also be absent, and squashed commits may not correspond one-to-one with the user's original commits.

Troubleshooting

• Install GitHub CLI and run gh auth login before using the dashboard.
• Check repository names use the owner/name form.
• If GitHub returns a rate-limit or permission error, check gh auth status and make sure the authenticated account can access the selected repositories.
• Run iseewhatyoudid --debug-info when reporting an environment issue.

Sponsors