Honestweek — Git skill for Claude Code
Find, replay and check your Claude Code and Codex sessions on your own machine: the sessions behind any pull request, commit or file, with every step tagged by how it's known.
How to install Honestweek
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open BryceEWatson/honestweek and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Honestweek does
Find, replay and check your Claude Code and Codex sessions on your own machine: the sessions behind any pull request, commit or file, with every step tagged by how it's known. It also writes an honest weekly summary checked against git. Local only, no telemetry.
Alternatives in Git
- Operator OSS — Run many Claude Code (or Codex) sessions in parallel — across every project — from one screen 202 ★
- Lore — git memory for coding agents — a local, searchable archive of your Claude Code and Codex sessions 142 ★
- Mimi Remote — Open-source native iPhone/iPad client for OpenAI Codex CLI and Claude Code — review diffs, approve actions, st 92 ★
README
honestweek
[](https://github.com/BryceEWatson/honestweek/actions/workflows/ci.yml) [](LICENSE)
See exactly where your Claude Code and Codex sessions went wrong, then browse every step, prompt and token behind it, and fix it. All on your machine.
honestweek works with the session logs that Claude Code and Codex already keep on your computer. It does three things with them, all on your own machine:
- See where it went wrong.
honestweek viewopens on a Problems page: the known ways AI coding agents go wrong that showed up in your sessions, claims the agent couldn't back first ("done" with no check after the last edit, or success the output doesn't show). Each one links to the step in your replay, says how it's known, shows whether it happened less than in the week before, and offers a fix you can copy into your own instructions or hooks. - Find and replay your work. Its Find page, one click away, takes a pull request number, a commit, a file, a branch or a few words and shows the sessions behind it, then lets you replay any session step by step. Every link and count says how it's known: recorded in a log or by git, computed from records, inferred by a named rule, or missing.
- Write an honest weekly summary. A short pipeline turns a finished week into a summary you review and publish yourself. It checks every commit the summary cites against your real git history first, and it stops rather than write a claim it can't back.
It's for developers who do much of their work through an AI coding agent and want to find, check or show that work later. Nothing leaves your machine: there's no account, no telemetry and no network call. The names and client words you list stay hidden on the page unless you turn on its Show private text switch, on your own screen, and keys, tokens and passwords stay hidden either w
Related Skills
Impl
EXECUTE PLAN — reads an existing implementation plan and runs all steps autonomously, verifies each step, writ
Architect Flow
Run the flow stage. Writes the user flows with a failure branch and a named human step each, plus the experien
Find Unfinished
Find work left half-done by interrupted agent sessions — reconstruct intent from recent session transcripts an
Laneboard
Shows all Claude Code sessions on one machine on a single page, grouped into lanes (one worktree, one branch,
Cc Remote
Use cc_execute/cc_check to run tasks on the user's local machine via Claude Code. Triggers when user asks to r
Claude Plan Skills
Claude Code skills for managing multi-step tasks as GitHub issues. Create, implement, update, and close execut
Related Agents
Ade
Agent sessions routinely outlast the machine that started them, and checking a Claude Code or Codex run from a
Text Slop Cleaner
This agent should be invoked to remove machine sounding text and unnecessary comments. This includes rewriting
Commit
Stage and commit changes, or open a pull request. Trigger when the user asks to commit, make a commit, save ch