Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/dork-labs/dorkos/context-isolatorgit clone --depth 1 https://github.com/dork-labs/dorkosWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/dork-labs/dorkos/context-isolator)<a href="https://agentmods.dev/agents/dork-labs/dorkos/context-isolator"><img src="https://agentmods.dev/badge/agents/dork-labs/dorkos/context-isolator.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00076 | $0.02592 |
| Opus 5 | $0.00038 | $0.01296 |
| Sonnet 5 | $0.00015 | $0.00518 |
| Haiku 4.5 | $0.00008 | $0.00259 |
Grade A, and why
context-isolator scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 229 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Context Isolator Agent
A lightweight read-only subagent for running operations that might return too much data for the main conversation context. Runs in a fresh context window, does the bloaty work, and returns a small structured summary.
Purpose
Protect the main conversation's context window by isolating data-heavy operations. The main context delegates the bloaty read + classify + format step to this agent, which does all the data-reading inside its own context window and returns only the condensed result.
This is the context-bloat escape hatch — one of the three legitimate reasons to spawn a subagent (the other two being tool-permission restriction and parallel execution).
When to Use
Use this agent when:
- Commit log analysis —
git logover many commits where you need a structured summary (counts by type, breaking markers, version recommendation). - Changelog parsing — reading the fragments in
changelog/unreleased/and classifying their entries. - Filesystem scans — grep/glob across many files where the raw hit list would be noisy but a structured summary is useful.
- Log aggregation — reading long server/build logs and extracting error patterns.
- Schema-diff classification — reading a multi-hunk diff and classifying each hunk (e.g., added field, removed field, type change) without bringing the full diff into main context.
- Unknown-result-size queries — any read operation where you're unsure how much data will come back.
When NOT to Use
- Simple targeted reads — reading a single file by path, grepping for one specific symbol you already know exists.
- Writes or mutations — this agent is read-only. Any task that edits files, makes commits, installs packages, or mutates external state belongs in the main context where the user can supervise.
- Small inputs — if the raw input is under ~50 lines, just read it directly in main context. The subagent has ~5s of spawn overhead that isn't worth it for small reads.
- Interactive workflows — tasks that need user confirmation mid-execution. The subagent returns once, so multi-turn back-and-forth is impossible.
- Tasks needing project-specific judgment — if the analysis depends on DorkOS architectural knowledge (FSD layers, Transport boundaries, ADR context), main context usually does better than a fresh subagent.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 229 lines · 76 tokens per session scan A 44c7aadbdc40
context-isolator is an agent published in the GitHub repository dork-labs/dorkos (9 stars, last pushed today), licensed MIT. It adds 76 tokens to every session and 2,592 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
multi-window
Agent "multi-window" from ThibautMelen/agentic-ai-systems, covering 🖥️ multi-window context, diagram, when to use, implementation and flow example.
updater
Processes .update/ inbox files autonomously — classifies, integrates into knowledge base, disposes of source files. Returns summary.
session-orchestrator
Session memory and continuity guardian managing state across context clears.
archivist
Memory investigation, synthesis, and authorized curation when surfaced memories are insufficient.
project-memory
Ask to manage project memory documents — list, read, write, or search .md files in the project memory directory.
sf-scribe
Records decisions and learnings to brain.db. Extracts patterns from completed work. Writes PR descriptions.