Install
$ agentstack add skill-webbertakken-agent-skills-session-reset ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ● Environment & secrets Used
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Session reset
Preserve session-specific state in PROMPT.md across a /clear, then restore afterwards.
PROMPT.md captures only session-specific state — current task, method, and progress. Persistent knowledge (user preferences, codebase patterns, dev quirks, workflow lessons) is handled separately by auto-memory and must NOT be included in PROMPT.md.
Workflow
1. Commit and push
Check git status for uncommitted changes. If there are any:
Security check — before staging, scan changed files for secrets. Flag and exclude any files matching common sensitive patterns (.env*, *.key, *.pem, credentials.*, *secret*, *.token). Warn the user if any are found and never stage them without explicit confirmation.
Then ask the user:
- Commit and push — stage safe changes, commit with a concise message, and push to the current branch
- Commit only — stage and commit without pushing
- Skip — leave changes as-is
If the user chooses to commit, use /commit. If they also want to push, push after the commit succeeds.
If the working tree is already clean, tell the user "Working tree is clean — nothing to commit." and move on.
2. Write PROMPT.md
Generate PROMPT.md at the project root by reviewing the current session state. PROMPT.md is ephemeral and must never be committed.
# PROMPT.md
## Current task
## Method
## Progress
Guidelines:
- Do NOT include user preferences, codebase patterns, dev quirks, or general knowledge — auto-memory handles that
- Do include task-specific context: current branch, active OpenSpec change, what's done, what's next, blocking issues
- Be specific and actionable — include file paths, branch names, change IDs, and concrete references
- Keep it concise — typically 20-40 lines
3. Tear down team (if active)
If a team is active (check for team config in $CONFIG_DIR/teams/):
- Read the team config to discover all active members
- List the active members to the user and confirm they want to tear down the team
- Send
shutdown_requestto each teammate - Wait for confirmations
- Delete the team with
TeamDelete
If no team is active, tell the user "No active team — skipping teardown." and move on.
4. Guide the user
After teardown, output this message exactly:
Session state saved to PROMPT.md. All agents shut down.
Run `/clear` now, then tell me: **read PROMPT.md**
5. Restore (post-clear)
When the user says "read PROMPT.md" after a /clear:
- Read PROMPT.md from the project root
- Summarise the restored context to the user
- Resume work based on the progress section
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: webbertakken
- Source: webbertakken/agent-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.