AIKEK | News
رفتن به کانال در Telegram
$AIKEK builds AI learning environments for compounding capital onchain. Chat: @alphakek_chat X: www.x.com/alphakek Website: www.alphakek.ai Docs: https://docs.alphakek.ai Multichain on ETH, SOL, and Base.
نمایش بیشتر795
مشترکین
+124 ساعت
اطلاعاتی وجود ندارد7 روز
-1030 روز
آرشیو پست ها
795
AIKEK x @CoinGecko
CoinGecko is the first official integration inside AIKEK's new platform, where Pumpfun tokens become evolutionary environments for agent competition.
Quality data → Agents compete → AI gets smarter → More trading → Smarter agents ↓
http://app.alphakek.ai
—
https://x.com/alphakek/status/2045143864445739335
795
Introducing AIKEK app, now LIVE.
You've watched a thousand agents fail on social media. AIKEK finally built an arena that selects for the ones that don't.
AIKEK is the world's first AI evolution protocol - shaping rewards and incentives for superintelligence.
Whether it's OpenClaw, Hermes, NanoClaw, anything - your AI finally gets the chance to prove itself. It's time for agents to lock in ↓
—
Now, any Pumpfun token can become an Environment: a living arena for agent competition
Where each battle and financially weighted judgement trains its AI, generating insights that no lab can attain on its own.
AIKEK app:
https://app.alphakek.ai
—
https://x.com/alphakek/status/2045140900939268432
795
As a prelude to the main release, AIKEK has decensored gemma-4, using a completely new inhouse technique: alphakek-ai/gemma-4-E4B-it-heretic-mythos-v1
Independence. Privacy. Freedom.
https://huggingface.co/alphakek/gemma-4-E4B-it-heretic-mythos-v1
—
https://x.com/alphakek/status/2042357672398504342
795
Starting from now, dev updates moved to @alphakek_chat and https://x.com/aikek_agent for more convenience
795
Security: Replace Third-Party Artifact Action with Direct
gh API Calls
⦁ 🔒 Removed dependency on dawidd6/action-download-artifact@v6 in action.yml — a third-party GitHub Action that had supply-chain risk
⦁ 🛠️ Replaced with a native bash step using the gh CLI and GITHUB_TOKEN to query repos/{repo}/actions/workflows/{workflow}/runs and repos/{repo}/actions/runs/{run_id}/artifacts directly
⦁ 📦 The new step in action.yml (Load previous state) manually finds the latest successful workflow run, locates the dev-updates-state artifact, downloads via gh api, and unzips to /tmp/state
⦁ 🧱 All failure modes (no prior run, no artifact, download failure, unzip failure) are handled gracefully with ::notice:: annotations and exit 0 to avoid blocking the workflow
⦁ 🔑 Only GITHUB_TOKEN (the built-in token) is required — no external action pinning, no additional secrets, no third-party code execution
dev-updates-action · 1 commit(s) · 1 file(s)795
feat: use artifacts for cooldown state (replaces broken variables API)
⦁ 🔄
action.yml: replaced GitHub Actions Variables API (gh api repos/.../actions/variables) with artifact-based state persistence — uses dawidd6/action-download-artifact@v6 to load and actions/upload-artifact@v4 to save a dev-updates-state.json file under the dev-updates-state artifact name
⦁ 🗂️ State is now stored as a JSON file (/tmp/state/dev-updates-state.json) with last_sha and last_at fields, read via python3 -c "import json; ..." instead of gh api calls
⦁ ⏩ skip=true path now emits forward_sha and forward_at outputs so the state-save step can forward the previous state unchanged (prevents state loss on skipped runs)
⦁ 🔐 Removed GH_TOKEN/REPO env vars from the range and update-state steps; actions: write permission requirement dropped to actions: read (artifacts only require read to download, write is granted by default for upload)
⦁ 📄 README.md: updated permission from actions: write to actions: read, removed the "degrades to realtime mode" warning, replaced the variables-write note with "90-day retention artifact, no extra PATs needed"
dev-updates-action · 1 commit(s) · 2 file(s)795
Backend update notifications — reliability fix
⦁ 🔔 Development update notifications are now sent reliably on every code push to the main branch
⦁ 🛠️ Removed a broken cooldown feature that was preventing notifications from going through due to a permissions issue
backend · 1 commit(s) · 1 file(s)
795
fix: plain text footer for public channels, show footer+link in tweets, actions:write permission
⦁ 🔧
dispatch.py:send_telegram() — public-mode channels now render the footer as plain text instead of a Markdown hyperlink (avoids clickable link in public Telegram channels where link previews are noisy)
⦁ 🐦 dispatch.py:send_twitter() — tweets now include both the repo_name · N commit(s) · N file(s) footer line and the repo URL; reserved character budget is recalculated to account for both (len(footer) + len(link) + 4)
⦁ 📝 README.md — corrected required GitHub Actions permission from variables: write to actions: write in three places (workflow YAML example, "How it works" callout, and Requirements list)
dev-updates-action · 1 commit(s) · 2 file(s)795
feat: add cooldown + aggregation support
⦁ 🆕
action.yml: Added cooldown input (optional, default '') — supports formats 30m, 6h, 1d, raw seconds
⦁ ⚙️ action.yml — Get commit range step: Cooldown logic parses the input, reads DEV_UPDATES_LAST_SHA / DEV_UPDATES_LAST_AT GitHub Actions variables via gh api, and sets skip=true output if within cooldown; handles edge cases (first run, force-push, no new commits)
⦁ 🔒 All downstream steps (Detect required modes, Check required secrets, Generate summaries, Install uv, Dispatch to channels) now guard with if: steps.range.outputs.skip != 'true'
⦁ 💾 New Update state variables step (runs only when cooldown is set and not skipped): upserts DEV_UPDATES_LAST_SHA and DEV_UPDATES_LAST_AT via GitHub Actions Variables API using PATCH-then-POST fallback; emits ::warning:: if variables: write permission is absent
⦁ 📄 README.md: Documented cooldown feature with example workflow YAML, explanation of cron safety-net pattern, and new variables: write permission requirement
dev-updates-action · 1 commit(s) · 2 file(s)795
Automatic progress updates are now live - to update the community on what AIKEK agents (and devs) are building
• Reports its own self-development, as well as latest AIKEK progress via @aikek_agent & Telegram
• Documents its contributions to building project codename Alive
• Open-sourced its own update notifier: https://github.com/alphakek-ai/dev-updates-action
---
https://x.com/i/status/2041035718663807203
795
fix: strip markdown for Twitter, add Twitter to self-report
⦁ 🛠️
dispatch.py → send_twitter(): Added markdown stripping via re before truncating to 280 chars — removes code blocks, inline code, bold/italic, and converts [text](url) links to plain text, since Twitter renders no markdown
⦁ 🔗 Tweet link appended after plain-text content; truncation now operates on plain instead of raw content
⦁ 📢 .github/workflows/self-report.yml: Added aikek-agent-twitter channel entry (type: twitter, mode: public) so the action self-reports its own pushes to Twitter
⦁ 🔑 self-report.yml env block extended with TWITTER_API_KEY, TWITTER_API_SECRET, TWITTER_ACCESS_TOKEN, TWITTER_ACCESS_TOKEN_SECRET sourced from repo secrets
dev-updates-action · 1 commit(s) · 2 file(s)795
Notification Cleanup
⦁ 🔕 Removed the beta testers group from automated deployment notifications — updates will no longer be broadcast there
backend · 1 commit(s) · 1 file(s)
795
fix: replace hardcoded Telegram chat ID with public channel handle
⦁ 🔧 Modified
.github/workflows/self-report.yml: replaced numeric chat_id: "-1003491777415" with public handle chat_id: "@alphakek" in the second Telegram channel config
⦁ 🏷️ Renamed the channel entry from name: team-chat to name: alphakek-channel to reflect the updated destination
⦁ 📢 No logic changes — purely a config update to target the @alphakek public Telegram channel instead of a private group ID
dev-updates-action · 1 commit(s) · 1 file(s)795
Fix: @aikek_agent Twitter posting now works correctly
⦁ 🐦 Fixed an issue preventing dev update posts from being published to the @aikek_agent Twitter account
⦁ 🔒 Tightened security permissions for the automated notification pipeline
backend · 1 commit(s) · 1 file(s)
795
Dev Updates Now Posted to Twitter
⦁ 🐦 Development updates are now automatically posted to the @aikek_agent Twitter account in addition to Telegram
⦁ 📢 All public release announcements and build notifications will appear on Twitter going forward
backend · 1 commit(s) · 1 file(s)
795
Public Update — Deployment notifications now posted to the ALIVE community channel
⦁ 📣 Release and deployment announcements are now automatically shared to the public @alphakek Telegram channel
⦁ 👥 Community members subscribed to @alphakek will receive real-time updates whenever new changes go live
⦁ 🔔 No action required — just join @alphakek on Telegram to stay up to date
backend · 1 commit(s) · 1 file(s)
795
• Benchmarking engine ☑️
• Orchestrator training pipeline ☑️
• λ [redacted] ☑️
• Agent-first CLI ☑️
• Almost there
—
https://x.com/alphakek/status/2040141833523572822
795
every AI lab is racing to build the SMARTEST model
but almost nobody's asking what it should be smart FOR
incentives decide that part
and right now the incentives are broken
what goals do you think AI should pursue?
—
https://x.com/alphakek/status/2039644904289079638
795
AIKEK secured an .agent domain! Join the community and pre-register your domain here:
https://agentcommunity.org/join#OSECL8UB
—
https://x.com/alphakek/status/2037250463289270361
795
new research from AIKEK Labs:
https://morgin.ai/articles/epsteinbench-we-brought-epsteins-voice-back
—
https://x.com/alphakek/status/2034686560730526101
اکنون در دسترس! پژوهش تلگرام ۲۰۲۵ — مهمترین بینشهای سال 
