Jump to the page content

Have an AI agent? Let it explore AgentFoundry for you. Any assistant works, personal or coding.

Connect your agent
What your agent reads

The request it copies

Explore getagentfoundry.si for me and report back. Read https://getagentfoundry.si/llms.txt, https://getagentfoundry.si/.well-known/agent-card.json and https://getagentfoundry.si/.well-known/mcp.json. Review the kits at https://getagentfoundry.si/kits and list the setup steps at https://getagentfoundry.si/connect. Stay read-only and check with me before installing any kit.

Discovery files on getagentfoundry.si

Read-only discovery is open. Keep AGENTFOUNDRY_KEY in your own environment.

AgentFoundry
BlogSign inBrowse the kits
Menu

Shared inbox kit

AF-EM-0110 filesMIT licence

Sorts a shared inbox, drafts replies to routine mail and flags what needs a person.

Install the kitIncluded in Builder, $19 a month

To install, sign in, or open a free account and add Builder.

Install the kitBuilder, $19 a month

Every part, in assembly order

MODELS.yaml is at revision B from October 22: its Claude Code block now sets effort for each step.

OrderPartWhat it doesTestRevision
1README.mdAF-EM-01-RMSetup, what success looks like and the recovery stepsSetup replays in a fresh projectA, Oct 18
2DEPENDENCIES.yamlAF-EM-01-DPMCP servers and libraries at exact versionsEvery pin resolves exactlyA, Oct 18
3MODELS.yamlAF-EM-01-MDModel, effort and output limit for each harnessOne block for each harnessB, Oct 22
4AGENTS.mdAF-EM-01-AGRole, boundaries and stop conditionsInside its 4,096 B budgetA, Oct 18
5RULES.mdAF-EM-01-RUNumbered rules, checked on every turnInside its 3,072 B budgetA, Oct 18
6SKILL.mdAF-EM-01-SKChecked methods, formulas and snippetsInside its 6,144 B budgetA, Oct 18
7TOOLS.jsonAF-EM-01-TLTyped tool schemas with path limitsEvery schema validates on loadA, Oct 18
8WORKFLOW.yamlAF-EM-01-WFOrdered steps, a call budget and recoveryRuns inside its call budgetA, Oct 18
9contract_card.jsonAF-EM-01-CCTyped ports, slot ID, refusal codes and checksumsEvery SHA-256 checksum matchesA, Oct 18
10verifier.pyAF-EM-01-VFIndependent checks, run apart from the workerReplays the recorded casesA, Oct 18

Standard files and size budgets: the AgentFoundry blog, October 18, 2026. Part numbers, tests and revisions are sample previews.

What it works with.

Checked with 4 models across 3 harnesses. verifier.py replays the kit’s recorded cases on each one before a release.

HarnessFiles it reads
Claude CodeCLAUDE.md, .mcp.json, .claude/settings.json
CodexAGENTS.md, ~/.codex/config.toml
OpenCodeAGENTS.md, opencode.json

Set it up in your harness.

The same ten files, mounted three ways. Your models, permissions and project stay yours. Or hand the job to your agent with this prompt.

Installed kit files carry the MIT licence: use, change and share them in your own projects. Builder installs kits over MCP, with setup for each harness and every new revision.

Agent prompt: check, then install
Read https://getagentfoundry.si/llms.txt. Get kit AF-EM-01, the shared inbox kit. Check each file against the checksums in contract_card.json, run verifier.py on its recorded cases and show me the result. Set it up in this harness after I say yes.

Claude Code

CLAUDE.md imports AGENTS.md and RULES.md, and a hook runs verifier.py after each edit.

# CLAUDE.md
@AGENTS.md @RULES.md
# .claude/settings.json
# the model block from MODELS.yaml
# PostToolUse hook, Edit|Write
python3 verifier.py

Codex

Codex reads AGENTS.md at the project root, and a WORKFLOW.yaml gate runs verifier.py.

# ~/.codex/config.toml
[mcp_servers.agentfoundry]
url = "https://getagentfoundry.si/mcp"
# project root
AGENTS.md  WORKFLOW.yaml

OpenCode

OpenCode reads AGENTS.md. opencode.json adds the server and the model from MODELS.yaml.

// opencode.json
"mcp": { "agentfoundry": {
  "type": "remote",
  "url": "https://getagentfoundry.si/mcp" } }