Alignbase

Self-improving, portable context for AI agents.

[Alignbase @alignbase_ai]

@AbeWheeler5 @NotionHQ https://t.co/6Iw3NufQmI

[Abe Wheeler @AbeWheeler5]

Finally, @alignbase_ai has a WYSIWYG editor!

@NotionHQ is the gold standard in editor experience and a major inspiration here.

We want our tool for managing your team's shared Skills, universal AGENTS[.]mds, and remote Memory to be as delightful as it is useful!

[Alignbase @alignbase_ai]

@AbeWheeler5 https://t.co/6Iw3NufQmI

[Abe Wheeler @AbeWheeler5]

Introducing @alignbase_ai Skill permissions & routing!

Now your context is always routed exclusively to agents that need it, with exactly the right access level.

No more dangerously exposed skills, missing change management, or un-auditable agent inputs!

[Abe Wheeler @AbeWheeler5]

Announcing shared, hosted memory for agents in @alignbase_ai!

Share your agent memories across models, harnesses, agents, and users.

Alignbase is now your headless, model-agnostic place to govern and coordinate:
🤖 Agents
🗂️ AGENTS.md
🔧 Skills
+ 🧠 Memory!

https://t.co/L6qIuP6K23

[Alignbase @alignbase_ai]

@AbeWheeler5 https://t.co/6Iw3NufQmI

[Abe Wheeler @AbeWheeler5]

Introducing fully-autonomous context!

On @alignbase_ai, agents proactively surface improvements to your shared AGENTS .mds, Skills, and context based on their work!

[Abe Wheeler @AbeWheeler5]

We just shipped agent context audits at @alignbase_ai! Now you can know exactly what your agents knew, every time they acted.

Now agent configuration is fully sharable, portable, and auditable! Bye bye, vendor lock-in 👋

[Abe Wheeler @AbeWheeler5]

@DavidSacks and @Jason on the last All-In talking about how nobody has solved context, skills, and memory portability between models… they haven’t seen https://t.co/wdc8yQq5Gd yet!

Bring everything enabling the intelligence layer with you, and make it self-improving 👀🇺🇸

[Abe Wheeler @AbeWheeler5]

Agent cold-start? I prefer "The Memento Problem".

Introducing Agent Proposals—let your agents store key learnings in team-wide AGENTS .md and Skill files. Agents don't forget, they improve the fleet.

@sunpeak_ai's Alignbase is better than tattoos.

[Abe Wheeler @AbeWheeler5]

Introducing the Alignbase Skills repository! With Skills, Alignbase is the place for teams to collaborate on ALL shared context, from the context your team's agents MUST use (like AGENTS .md) to the context they CAN use (Skills). @sunpeak_ai

[Abe Wheeler @AbeWheeler5]

Introducing Alignbase: your company-wide AGENTS.md

The engineers that used to write the code knew the company, product, and policies.

Now, the agents writing the code don’t. 🧵 (1/6)

[Abe Wheeler @AbeWheeler5]

Shipped @sunpeak_ai evals. One command tests whether LLMs actually call your MCP tools correctly, across GPT-4o, Claude, Gemini Flash, 4o-mini, etc.

Tools work on flagship models but silently break on the budget models users actually run.

100% on GPT-4o, 40% on Gemini Flash. That gap is a schema problem you only find out about from user complaints.

[Abe Wheeler @AbeWheeler5]

We just shipped @sunpeak_ai as a standalone testing framework for MCP Apps 🚀 Test everything about your MCP App, no matter where or how you built it!

- Unit tests (Vitest + happy-dom)
- E2E tests against replicated ChatGPT and Claude runtimes
- Visual regression testing with screenshot baselines
- Live tests against real ChatGPT
- Works with any MCP server in any language (Python, Go, TypeScript)