ds-context-reset — Claude Code skill: what it does, how to install
ds-context-reset is a Claude Code skill that manages context resets for long tasks using a PROGRESS.md tracking file.
MIT 562 installs ★ 320 updated 2026-10-28
npx skills add aicodeking/deepseek-stack --skill ds-context-reset -a claude-codeWhat ds-context-reset does
The skill stores work state in PROGRESS.md instead of the chat transcript so a fresh context can continue without drift. When the model starts repeating itself or the context window fills up, the skill writes a handoff file and moves execution to a new session or subagent.
The tracking file holds the goal, done steps, next actions, open questions, and key facts. It must stay under about 150 lines, and secrets or tokens are strictly forbidden inside.
Execution stops if three consecutive resets happen on the same next item without progress, prompting the user for intervention with the report file.
What it helps with
- store work state in PROGRESS.md for long tasks
- move execution to a fresh session or subagent
- reset stale or half full model context
- track verified steps and key facts
Example prompts for the agent
- /ds-context-reset
- Run /ds-context-reset to handle context drift
- Prepare a context handoff using /ds-context-reset
How to install ds-context-reset
1Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-context-reset -a claude-code2Manually: put the files in place
~/.claude/skills/ds-context-reset/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/ds-context-reset/
1Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-context-reset -a codex2Manually: put the files in place
.agents/skills/ds-context-reset/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-context-reset -a cursor2Manually: put the files in place
.cursor/skills/ds-context-reset/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install aicodeking/deepseek-stack skills/ds-context-reset2Manually: put the files in place
.github/skills/ds-context-reset/SKILL.md1Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-context-reset -a gemini-cli2Platform CLI
gemini skills install https://github.com/aicodeking/deepseek-stack --path skills/ds-context-resetManually: put the files in place
.devin/skills/ds-context-reset/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add aicodeking/deepseek-stack --skill ds-context-reset -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about ds-context-reset
How to install ds-context-reset in Claude Code?
Run npx skills add aicodeking/deepseek-stack --skill ds-context-reset -a claude-code, then check ~/.claude/skills/ds-context-reset/SKILL.md.
What does ds-context-reset do?
It records progress in PROGRESS.md when the context gets stale and helps transition execution to a fresh session.
What is needed to run ds-context-reset?
A repository root with a PROGRESS.md file and a fresh session or subagent to read the handoff.
Source text
SKILL.md · MIT
--- name: ds-context-reset description: "Keep a PROGRESS.md for long tasks and move to a fresh session or subagent that reads only that file when the context gets long or stale. Use when a task spans many steps or sessions, when the model starts repeating itself, forgets earlier facts, or re-tries old fixes, or when the context window is past about half full. Invoke as /ds-context-reset." --- # ds-context-reset Store the state of the work in a file, not in the chat, so a fresh context can continue without drift. ## Steps 1. At the start of a long task, create PROGRESS.md at the repo root with these sections: - **Goal**: one or two sentences, plus the acceptance check for the whole task. - **Done**: the verified steps, each with the command and result that proved it. - **Next**: the ordered steps that remain. The first item is the exact next action. - **Open questions**: unknowns and decisions that wait for the user. - **Key facts**: paths, commands, versions, gotchas, and approaches that failed (with the reason). 2. Update PROGRESS.md after every verified step. Do not update it for unverified ones. Commit the code and PROGRESS.md together with a descriptive message. 3. Watch for reset signals: you repeat an earlier answer, you forget a fact that is in Key facts, you retry a fix that failed, the same file gets reread many times, or the context is more than about half full. 4. When a signal appears, finish or revert the current step so the tree is in a known state. 5. Rewrite PROGRESS.md as a complete handoff. A reader who has only this file and the repo must be able to continue. Remove chatter and keep the facts. 6. Start a fresh session or subagent. Give it one instruction: "Read PROGRESS.md and `git log -5`, run the acceptance check for the last Done item, then do the first item in Next." 7. In the new context, confirm that the baseline check passes before you make new edits. If it fails, fix that first. ## Stop when - The goal's acceptance check passes and PROGRESS.md shows every item as Done. - You have done 3 resets on one Next item without progress. Stop and ask the user, with the file as your report. ## Rules - PROGRESS.md holds facts and evidence, not plans that nobody has checked. - Keep it under about 150 lines. Move old Done items into a short summary line. - Never put secrets or tokens in it. - The fresh context reads the file, not the old transcript. - Pair with ds-plan-loop for the Next list and with ds-unstick for the failed-approach notes.
© aicodeking · MIT · aicodeking/deepseek-stack