--- title: How Much MCP Output Is Waste? Measuring Context Pollution description: Clear-cut waste was 0.2% of the largest live request; unread records up to 12.5%. The exception: a feed where 152 of 200 events were the agent's own calls. canonical: https://spiritroom.app/blog/mcp-context-pollution/ published: 2026-10-21 updated: 2026-10-21 --- > Markdown version of https://spiritroom.app/blog/mcp-context-pollution/ for agents. Index: /llms.txt [Blog](/blog/) / [Tool results and errors](/guides/tool-results-and-errors-for-agents/) # How much of your MCP output is waste? Measuring context pollution in agent runs “Agents waste most of their context on tool output” is a common claim. We measured it with explicit rules and a hand check. The honest answer: clear-cut waste is small, the fuzzy kind is bigger, and one design mistake dominates. Leonid Bugaev · Published 21 Oct 2026 **TL;DR** - **Five clear-cut kinds:** identical repeat calls, errors over 1k tokens, refusal guidance, the agent's own logged calls, and hidden records the run didn't use. - **Small live:** 0.2% of the largest request in 40 live Codex runs. - **Except one feed:** in runs that read an activity feed, 7.1%, almost all the agent's own tool-call events. 152 of 200 events in one call were the connection's own calls. - **Unread records are the bigger, fuzzier share:** up to 12.5% (an upper bound; a hand check found the rule 62% precise). - **MCP results were 12–46%** of the largest request; the rest was definitions, search output and the app's prompt. [What counts as waste](#classes)[What we found](#results)[The feed full of footprints](#feed)[Unread records: the fuzzy share](#fuzzy)[How to measure yours](#measure)[How we measured](#how-we-measured)[FAQ](#faq) ## What counts as waste We classified every MCP result token that entered the model's context, first rule wins: 1. **Identical repeat:** same tool, same arguments, same result as before in the run. 2. **Oversized error:** an error over 1,000 tokens. 3. **Refusal guidance:** boilerplate telling the agent to stop and ask the user. 4. **Self-generated events:** the agent's own tool calls, logged back to it as activity. 5. **Hidden-stage records** the run didn't use. 6. **Unreferenced records:** records whose distinctive values (IDs, names, numbers) never appear in the answer or any later model text or tool call. Classes 1–5 are **strict** waste. Class 6 is fuzzy, so strict + 6 is an **upper bound**. ## What we found Live Codex, 8 read jobs: strict *0.2%* Live Codex, 8 read jobs: upper bound *12.5%* Live runs reading the feed: strict *7.1%* Live runs reading the feed: upper bound *19.1%* Share of the largest single request, averaged per run (bars scaled to 50%). In cumulative input the live figures were 0.1% strict and 9.0% upper. MCP results themselves were 12–46% of the largest request; see [MCP context window usage](/blog/mcp-context-window-usage/). **Measure the waste in your MCP output.** We classify every token your server returns in real agent runs and show which tools and fields to cut. [Measure your MCP →](/#contact) ## The feed full of footprints The server's “what happened since” tool logged every MCP tool call as an event. With a limit of 200, 200 events came back and **152 were this same connection's own tool calls**. The response was 98,512 tokens. In the runs that read it, those events were about 4k tokens per run, almost all of the 7.1% strict waste. Fix: leave the caller's own tool-call events out by default, behind a filter. ## Unread records: the fuzzy share Typical cases: 49 other records returned when the user asked about one, and long strategy blocks nobody read. But string matching misses paraphrase, counting and summing. A hand check of 5 runs found the rule **62% precise**, overstating the upper bound by 6.9 points on average, and just 1% precise on a summarising job. Scaled by that, live unread waste is roughly 8%: an estimate, not a measurement. Fixes are the same as for size: filters, small defaults, summaries. See [reduce tool response size](/blog/reduce-mcp-tool-response-size/). ## How to measure yours - Record MCP results as the model saw them, after any app truncation. - Apply the strict rules; report strict and upper bound separately. - Hand-check a few runs, including a summarising job, before trusting the upper bound. - Look for duplicates: identical repeats are rare, but repeat calls with new arguments aren't waste. See [why agents call the same tool twice](/blog/duplicate-tool-calls/). **Go deeper:** [The 12,689-token typo](/blog/verbose-validation-errors-tokens/), [Agent refusal states](/blog/agent-refusal-states/), [Per-task cost, Claude Code vs Codex](/blog/mcp-task-token-cost-claude-code-vs-codex/). Back to [tool results and errors for agents](/guides/tool-results-and-errors-for-agents/). ## How we measured - **Live:** 40 Codex CLI 0.161.0 runs of 8 read jobs plus 10 earlier runs of 2 jobs, on a 99-tool production MCP server through a read-only proxy, 16 Oct 2026; tokens by o200k\_base. - **Mock:** 150 runs on a live-shaped mock, for comparison. - **Hand check:** 5 runs, every flagged record judged needed or not; strict rules agreed on all 5 (in-sample). False negatives weren't measured systematically. - **Limits:** one workspace, read-heavy jobs, one day. ## FAQ ### How do I measure wasted tokens in agent tool results? Split every MCP result that enters the model's context into classes with explicit rules: identical repeats, oversized errors, refusal guidance, self-generated events, and records never referenced later. Report the clear-cut classes as strict waste and unreferenced records only as an upper bound, and hand-check a few runs. ### Is most MCP output wasted? Not in our runs. Clear-cut waste was 0.2% of the largest request in live runs. Unreferenced records reached 12.5%, but about 38% of those were actually used through paraphrase, counts or sums. ### What was the biggest source of waste? An activity feed that logged the agent's own tool calls as events. In one call 152 of 200 events were the connection's own calls; in runs that read it, strict waste rose to 7.1%. **Measure the waste in your MCP output.** We classify every token your server returns in real agent runs and show which tools and fields to cut. [Measure your MCP →](/#contact) ## Keep reading - Up[Tool results and errors for AI agents (guide)](/guides/tool-results-and-errors-for-agents/) - Related[Why your MCP server fills the context window, and which agents actually pay for it](/blog/mcp-context-window-usage/) - Related[How to reduce MCP tool response size: measured before and after](/blog/reduce-mcp-tool-response-size/) - Related[Why agents call the same MCP tool twice (13 of 40 runs) and how to stop it](/blog/duplicate-tool-calls/)