Your company runs on
what a few people remember.
Lore reads everything your team writes across Slack, GitHub, Linear, Gmail, PagerDuty, and Notion, and emits executable SKILL.md files your AI agents load and act on. The brain stays current as the company learns.
Demo workspace, three Skills, real agent runs. No signup required beyond the demo password.
When this Skill applies
A customer requests a refund or cancellation.
Decision criteria
Edge cases
Hi team, we removed two Pro seats on May 8 and would like a prorated refund. Comes out to roughly $480. Account is on monthly billing. Thanks, David.
- 01Customer is Acme Logistics, requesting $480 prorated refund for unused seats.obs:a3f1c2
- 02$480 < $500 threshold. Routes to CSM-tier auto-approve.obs:a3f1c2
- 03No edge cases triggered (account is monthly, not annual).obs:e4c8d1
- 04Drafting a customer reply addressed to David.obs:a3f1c2
Hi David, thanks for flagging this. We've approved the $480 prorated refund for the two unused Pro seats and it'll land back on your card within 3 business days. Let us know if anything else looks off. Maria, Northstar CSM
Reply to ops@acme-logistics.com (David). Awaiting Maria's approval before send.
Connects to the tools your team already lives in
The primitive
A company brain, not a search index.
The right way to tell Lore apart from search is to look at what it produces. Procedures, not links. Agents act on them.
What it is
Reads everywhere work happens
Slack, GitHub, Linear, Gmail, PagerDuty, Notion. The brain sees what your team writes, not what they remember to document.
Compiles work into playbooks
When the same decision happens three times, Lore writes it down as a SKILL.md file. Versioned. Cited back to the source. Loaded by agents via MCP.
Stays current automatically
New observations stream in, the playbook re-emits, the version bumps. The brain ages with the company instead of going stale in someone's Notion.
What it is not
Not a wiki
We don't store the docs you write. We read the work that's actually happening on top of them.
Not a company-wide search
Search returns links. Lore emits procedures an agent can execute, with every claim cited to a source observation.
The audit loop
Every run, logged.
When an agent loads a Skill or runs it on a ticket, the call lands here. Version, agent, outcome, duration. Failed runs flag the Skill for the next refresh.
Runs . incident-response
Every time an agent loads or applies this Skill, one row lands here. Failed runs surface as candidates for Skill improvement.
| When | Agent | Kind | Ver | Dur | Status |
|---|---|---|---|---|---|
| just now | Claude Code | execute | v1.0.4 | 5.2s | success |
| 3m ago | Cursor | load | v1.0.4 | 0.1s | success |
| 12m ago | Lore dashboard | execute | v1.0.4 | 6.8s | success |
| 1h ago | Claude Code | execute | v1.0.3 | 32.0s | failed |
| 2h ago | Windsurf | load | v1.0.3 | 0.1s | success |
| 3h ago | Claude Code | execute | v1.0.3 | 31.5s | failed |
| 4h ago | Codex CLI | load | v1.0.3 | 0.1s | success |
| 6h ago | Claude Code | execute | v1.0.3 | 5.9s | success |
Built to scale
The graph keeps everything. The file an agent loads stays small.
Hot sections by default, cold sections archived for audit, redundant observations distilled to canonical rules. As the company grows, the file an agent pays context for stays lean.
Refund handling
v1.0.4Authority matrix and edge cases for processing customer refunds.
Manifest
~548 hot tokens . ~666 full5 hot sections loaded by default for agents via MCP. 1 cold section included only when explicitly requested.
21 observations distill to 5 canonical rules. Citations on the body link back to the supporting observation cluster.
Connect your team.
See your brain emerge.
Ten design partner spots. Connect Slack, GitHub, Linear, and email; the first three Skills land in an afternoon.