FailEcho is a cross-agent failure intelligence network. Agents check it
before retrying a failed tool call to see whether other agents are hitting
the same failure and which recovery worked.
Placed alphabetically in the Monitoring section. Python codebase, available
both as a hosted remote server and self-hosted.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QzYv4T5Hg6fiU1sAkGpYKx
Two MCP servers for color-vision accessibility: palette/image colorblind-safety
checking with the conflicting pairs named, colorblind-safe palette generation,
image recoloring for protan/deutan/tritan, and Ishihara-style plate generation.
Machado, Oliveira & Fernandes (2009) + CIEDE2000; method published open access
at doi.org/10.5281/zenodo.21310578. Local stdio, no network calls.
Drop2Run MCP server: publishes static output to a live HTTPS URL. Three tools — publish_files, publish_dir, list_sites. Published on npm as @drop2run/mcp, MIT, public repository, run by the user via npx.
The entry described the server as a "security gateway with tool-access
policies"; the README, the site and the Glama listing all lead with
"the policy enforcement plane for MCP". Both are true, but a directory
carrying retired positioning is how copy drifts out of sync in the
first place.
Now byte-identical to README.md line 5, minus the bold markers this list
does not use.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016o7mBbWapnVqDAPRJb4z1K
The badge goes after the repository link and before the emoji tags,
matching the neighbouring entries, rather than after the description as
the bot's instructions word it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016o7mBbWapnVqDAPRJb4z1K
The bot tagged this PR missing-glama. The server is now listed on Glama(glama.ai/mcp/servers/xsf0221/pattern-vista-cli), so the badge URL that the other entries in this section carry now resolves for it too.
Two corrections to the pending entry, both mine.
The corpus line said the texts were written 'before generative models were in
use'. GPT-3's API was in beta from mid-2020, so that is not true of a 2021
revision; the claim that holds is 'before 2022', which is weaker and correct.
It is the wording the project itself uses in Docs/CALIBRATION.md.
And the entry counted tools -- 'ten in all', 'eight of the ten'. An entry that
enumerates goes stale the next time one ships; one that characterises does not.
This now says 'every tool but two runs entirely on the machine', which stays
true without anybody editing it.
Also adds the install command, since the package is on NuGet.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015PEbbiYSNPw7jE3LrPNhyF
Requested by the glama-check bot (2026-07-16) and followed up by @punkpeye
(2026-07-29). The server is listed on Glama and the quality score is now set
(rated B). Badge placed after the repo link, before the platform flags, to
match the existing entries in this file.
dockndevai
omnom62
Fuyuki0andClaude Opus 5
fengbs600220
ami-guru
joeygambino
sikcapri
pepinuandClaude Fable 5.1
Yunfeng Bai
zzzjy765
Abhishek Kandivalikar
Jake Barnby
Christian Velasquez
netanelavr
dahshanlabs
ykforerlang
flightmasterDast
Khanhz
Vince Gonzalez
Marcin PyrkaandClaude Opus 5
webmilmind1
Timur Rakhmatullin
sf.Hsing
Pedro HernándezandClaude Opus 5
stillmarcus24
WhiteMagic AI
Prompeteer
kleinicke
Dominic BussemasandClaude Opus 5
Sensie-agents
Prateek Singh
Moby MediaandClaude Opus 5
Abhishek Kandivalikar
Alper San