kodelyth-eccBLOCK
70 agents · 196 skills · 103 commands · 44 hook entries · semantic intent routing · self-learning memory · GOD vs EVIL arena · MCP server. macOS, Linux and Windows. Claude Code, Windsurf, Cursor, Codex, Antigravity, Cline, Roo Code, Aider, Kimi, Gemini CLI. MIT.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
Describe your problem in plain words. The right specialist shows up.
No agent names to memorise. No prompt engineering. Nothing leaves your machine.
You: "I've been staring at this NullPointerException for two hours, I'm losing my mind." AI: → Routing to debug-detective (your error + frustration matches the bug-tracking signal) That kind of bug is exhausting — let's trace it properly so we fix the root cause, not the symptom. First, can you share the full stack trace and...
You never typed use debug-detective. You did not have to — ECC read the intent, picked the specialist out of 70, and told you which one it chose and why.
npx kodelyth-ecc
Works on macOS, Linux and Windows with Claude Code, Windsurf, Cursor, Codex CLI, Google Antigravity, OpenCode, Cline, Roo Code, Aider, Kimi and Gemini CLI.
[](https://www.npmjs.com/package/kodelyth-ecc) [](https://www.npmjs.com/package/kodelyth-ecc) [](https://github.com/sifxprime/kodelyth-ecc/stargazers) [](LICENSE) ![Code
3a56024d24ebOBSERVED · 2026-10-12Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add openclaw-twitter --env ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} --env GITHUB_TOKEN=${GITHUB_TOKEN} --env KODELYTH_TOKEN_BUDGET=${KODELYTH_TOKEN_BUDGET} --env KODELYTH_TOKEN_BUDGET_DIR=${KODELYTH_TOKEN_BUDGET_DIR} -- npx -y [email protected]{
"mcpServers": {
"openclaw-twitter": {
"command": "npx",
"args": [
"-y",
"[email protected]"
],
"env": {
"ANTHROPIC_API_KEY": "${ANTHROPIC_API_KEY}",
"GITHUB_TOKEN": "${GITHUB_TOKEN}",
"KODELYTH_TOKEN_BUDGET": "${KODELYTH_TOKEN_BUDGET}",
"KODELYTH_TOKEN_BUDGET_DIR": "${KODELYTH_TOKEN_BUDGET_DIR}"
}
}
}
}Exposed tools (22)
22 read · 0 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
agents-overview | read | A compact overview of every ECC agent (name + one-line description). Useful for orientation. |
audit_skill_match | read | Suggest ECC skills whose description/body overlap a task. Use to decide which skills to attach as context. |
capture_memory | read | Append a new memory entry to the local Kodelyth memory store. |
catalog_stats | read | Summary of how many ECC agents, skills, commands, rules, and bundles are loaded. |
commands-overview | read | A compact overview of every ECC slash command. |
devil-mode | read | The /devil-mode adversarial parallel command — fires the red-team crew at a target. |
get_agent | read | Fetch the full markdown body of a single agent. |
get_bundle | read | Fetch the cheat sheet for a power bundle (indie-hacker, red-team, enterprise). |
get_command | read | Fetch the full markdown body of a single slash command. |
get_rule | read | Fetch the full markdown body of a single ECC rule file. |
get_skill | read | Fetch the full markdown body of a single skill. |
handoff-chains | read | Standard ECC multi-agent handoff chains for common workflows (new feature, bug fix, refactor, incident, etc.). |
list_agents | read | List all ECC agents (name, description, relative path). |
list_bundles | read | List ECC power bundles (indie-hacker, red-team, enterprise). |
list_commands | read | List all ECC slash commands. |
list_rules | read | List all ECC rule files in rules/common/. |
list_skills | read | List all ECC skills (name, description, origin, relative path). |
memory_stats | read | Summary of the local Kodelyth memory store (counts, projects, etc.). |
recall_memory | read | BM25 search across the local Kodelyth memory store. Returns top-N matching memories. |
route_intent | read | Suggest the best ECC agent for a user message via token-overlap scoring. Read the full routing rule via the |
routing-rule | read | The full ECC intent routing rule — 10-tier priority system mapping user intent to specialist agents. |
skills-overview | read | A compact overview of every ECC skill (name + one-line description). |
Trust audit
BLOCKgrade F · trust 35/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | FAIL |
| L2 | Instruction surface (what it tells the agent) | FAIL |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (6 observation(s))
- Network
- declared (12 observation(s))
- Shell
- declared (9 observation(s))
- Dependencies
- not all pinned
- Secrets in source
- found
Findings (25)
// Whitelist only the env vars the server needs — never forward all of process.env
user = User.objects.create_user(email='[email protected]', password='testpass123')
const result = exec(resolved.bin, args, projectRoot);
const result = exec(resolved.bin, args, projectRoot);
exec(sql) {3. Scan for risky patterns (\`eval(\`, \`dangerouslySetInnerHTML\`, unsanitized \`innerHTML\`, obvious SQL string interpolation).
/id_rsa/,
DATABASE_URL: postgres://postgres:postgres@localhost:5432/test
- DATABASE_URL=postgres://postgres:postgres@db:5432/app_dev
postgres://postgres:postgres@db:5432/app_dev # "db" resolves to the db container
export JIRA_API_TOKEN=$(op read "op://Private/Jira/token") # 1Password
1. Read `.env.example` (or `.env.template`, `.env.sample`)
Before running any VideoDB code, change to the project directory and load environment variables:
**Do NOT** read, write, or handle the API key yourself. Always let the user set it.
| "Rebase has conflicts I don't understand" | Show the user the 3 versions (ours/theirs/base), explain who wrote what |
explicitly guard it — do not delete the test, and do not tell the user "all tests
## v2.6.0 — GOD mode + EVIL mode v2 (Arena phases 0-2) (August 2026)
- **Adversarial verification** — each top finding is challenged by a fresh agent told to **REFUTE it**, defaulting to refuted when uncertain. False positives die before they can waste GOD mode's next
### Added — Phase 2: GOD mode (`scripts/arena/god.js`, `/god-mode`)
- **`hooks/safety/prompt-injection-guard.js`** — scans inbound text on `UserPromptSubmit` and `PostToolUse` (Read, WebFetch, mcp__*) for jailbreak / instruction-override patterns. Three modes: `off` (
- **high** — hidden system markers (`[[SYSTEM]]`, `<|system|>`, `### NEW INSTRUCTIONS`), tool-call hijack patterns, jailbreak canaries, demands for unrestricted output
id: 'exfiltration-channel',
regex: /\b(send|post|upload|exfiltrate|leak|forward)\s+(?:the\s+)?(secrets?|api[\s_-]?keys?|credentials?|env(?:ironment)?\s*(?:vars?|variables?)?|tokens?|sessions?|cookies?|chat\s+history|conversation
why: 'Tries to direct the model to exfiltrate sensitive data.',
regex: /\b(call|invoke|run|execute)\s+(?:the\s+)?(tool|function|mcp|command|shell)\s+["'`]?(rm|curl|wget|fetch|bash|sh|exec|eval|delete_|drop_|exfil_|send_|post_|leak_)/i,
Gates applied: critical_finding, instruction_override, no_behavioural_pass, undeclared_transfer.
3a56024d24ebfull audit observations/trust-audit/mcp-server/sifxprime__kodelyth-ecc.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-12 | 3a56024d24eb | BLOCK | F | 35 | first audit |
Questions
What is the kodelyth-ecc MCP server?
70 agents · 196 skills · 103 commands · 44 hook entries · semantic intent routing · self-learning memory · GOD vs EVIL arena · MCP server. macOS, Linux and Windows. Claude Code, Windsurf, Cursor, Codex, Antigravity, Cline, Roo Code, Aider, Kimi, Gemini CLI. MIT.
What tools does kodelyth-ecc expose?
22 in total: 22 read-only, 0 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is kodelyth-ecc safe to connect to an agent?
No — not without reading the findings first. The audit graded it F (35/100) and found 21 critical or high issues in the source. Each one is listed on this page with the file and line it is on.
What credentials does kodelyth-ecc need?
It reads ANTHROPIC_API_KEY, GITHUB_TOKEN, KODELYTH_TOKEN_BUDGET, KODELYTH_TOKEN_BUDGET_DIR, KODELYTH_TOKEN_BUDGET_RESET and KODELYTH_TOKEN_BUDGET_WARN from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does kodelyth-ecc run?
It speaks stdio, so it runs as a local process your client starts. It is published on npm as openclaw-twitter at 1.0.0.
How current is this page?
The grade is for one exact copy of the source (3a56024d24eb), read on 2026-10-12. The repository is watched and re-audited when it changes.