A like is a signal in the decision ledger, reachable the same way for a person and an agent, and a test reads it back #10389

One sentence: the feed's like must be a ledger entry any caller can write and read, not a button's side effect.

Done:

• [x] Liking a post from the page, from the API and from the MCP writes one ledger row (who, what, when) and is idempotent: a second like is a no-op that answers the same.
• [x] Unlike removes it by a new ledger row, never by editing history.
• [x] The like count on a post equals the ledger's count for it; the test asserts this after a like, unlike and repeat.
• [x] A like needs a same-origin POST and a session, or a token with the right scope; an anonymous caller gets 401 with the sign-in route named.
• [x] docs/FEED.md says where a like is stored and how to read it.

Files: src/feed/feed.ts, src/actions/ (the like action), src/http/mcp/tools/feed.ts, test/feed-like.test.ts, docs/FEED.md Risks: like-flooding by an agent; reuse the existing per-caller rate limit and assert it. Suggested: scout.

Reviewed pull #10410 at its head: ticked Liking a post from the page, from the API and from the MCP writes one ledger row (who, what, when) and is idempotent: a second like is a no-op that answers the same.; Unlike removes it by a new ledger row, never by editing history.; The like count on a post equals the ledger's count for it; the test asserts this after a like, unlike and repeat.; A like needs a same-origin POST and a session, or a token with the right scope; an anonymous caller gets 401 with the sign-in route named.; docs/FEED.md says where a like is stored and how to read it..