Context Guard
Context window protection. Auto-activates when the conversation is getting long, when the user is about to do a large file read or exploration task, or when the context is approaching its limit. Helps route expensive operations to sub-agents to protect the main session.
Install
$ agentstack add skill-shivae372-claude-bootstrap-context-guard ✓ 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 No
- ✓ 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.
Links to your public security report.[](https://agentstack.voostack.com/security/report/skill-shivae372-claude-bootstrap-context-guard) 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
Purpose
Protect the user's context window from expensive operations that should run in a sub-agent instead. This skill activates automatically when it detects patterns that commonly cause context blowout.
When I Activate
- User asks to "read", "scan", "explore", or "analyze" a large codebase
- User asks to run tests and see all output
- Context window is getting long and a compaction is likely soon
- User is about to do something that will dump large amounts of text into the main context
What I Do
1. Assess the Risk
Before the expensive operation happens, estimate the token cost:
- Reading a large file: HIGH risk
- Exploring a directory: MEDIUM-HIGH risk
- Running tests with verbose output: HIGH risk
- A targeted grep: LOW risk
2. Route to Sub-agent If High Risk
If the operation is HIGH risk, do NOT execute it in the main context. Instead:
I'm routing this to the [explorer/test-runner] sub-agent to keep our main context clean.
Use: "[agent-name] — [what you need]"3. Compress Before Returning
If a sub-agent result is returned to you, summarize it before adding it to the main conversation:
- Max 20 lines for any exploration result
- Max 10 lines for a test run summary
- Always structured (JSON or bullet list), never raw file dumps
4. Warn When Context Is Large
If the current context is approaching the limit:
⚠️ Context Warning: Our session is getting long. I recommend:
1. Using /compact to compress history
2. Starting a new session for the next distinct task
3. Current state will be saved to SESSION_STATE.md firstWhat I Do NOT Do
- Block the user from doing anything
- Make the decision unilaterally — always explain and ask if the rerouting is okay
- Obsessively activate on every message (only on genuinely expensive operations)
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: shivae372
- Source: shivae372/claude-bootstrap
- License: Apache-2.0
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.